Senate Banking Committee holds hearing on artificial intelligence

[0:00] companies are addicted to debt. [0:02] Early in AI development, these companies [0:05] financed their own investments. [0:07] But now, they are borrowing upwards of $1 trillion [0:12] to finance spending on data centers, chips, [0:16] and other AI infrastructure. [0:18] They borrow from private credit funds that, in turn, [0:23] borrow from banks. [0:24] This means that if AI tumbles, it [0:28] will bring down many in private credit, [0:31] and that, in turn, will risk bringing down giant banks. [0:36] The possible failure of AI companies [0:39] looms larger every day. [0:42] AI companies that are unable to increase revenues and service [0:45] their massive debt loads could trigger massive losses [0:50] in the financial sector and another 2008-style crash. [0:55] That, in turn, could wipe out huge amounts of value [0:59] from Americans' retirement accounts. [1:02] Third, Americans are worried about what [1:04] will happen to their jobs in a world of rapidly improving AI [1:09] capabilities. [1:11] AI has already started reshaping the workforce. [1:14] While we don't know what the long-term effects on unemployment [1:19] will be, we cannot wait around to see how many jobs will [1:24] be eliminated before we start to prepare for the coming upheaval. [1:29] We have a decision to make. [1:32] Either AI will profit a handful of the richest people, [1:37] making them even richer, or the benefits of AI [1:41] will be used for all Americans. [1:44] If we continue to do nothing, the billionaires [1:47] will suck up all the value from AI. [1:50] But if we start now to tax AI and to use that money [1:55] to fund free education, apprenticeships, [1:58] and encourage new job creation for the workforce, [2:01] we can share the benefits of AI across our country. [2:07] Before we hear from our witnesses, [2:09] I want to raise one other issue. [2:12] China is also working feverishly to develop AI, [2:15] including AI weapons systems that [2:18] can be turned against the United States. [2:22] One company, NVIDIA, makes the chips that [2:26] are powering AI's advances. [2:29] In recent years, NVIDIA's valuation [2:32] has ballooned to nearly $5 trillion. [2:35] That makes NVIDIA the world's biggest company, [2:39] and NVIDIA's CEO, Jensen Huang, a very powerful person. [2:45] And Mr. Huang wants to sell NVIDIA's chips to China, [2:50] which will profit his own company and advance China's [2:54] AI development. [2:56] Now, I invited NVIDIA's CEO, Jensen Huang, [2:59] to testify at this hearing, and he refused to join us. [3:03] I invited Mr. Huang because he and NVIDIA [3:07] are lobbying hard to get a blank check from the Trump [3:11] administration to allow the sale of advanced chips to China. [3:16] Mr. Huang's efforts have alarmed senior military leaders here [3:20] in the United States who have warned that such technology could [3:25] strengthen China's military and put our own national security at risk. [3:32] No matter. In Trump's D.C., money talks, and President Trump appears [3:38] to be giving policy handouts to the highest bidder. [3:43] If it's not plain old-fashioned bribery, it sure smells like it. [3:48] Mr. Huang and NVIDIA have been active here in D.C. [3:52] NVIDIA donated a million dollars to the Trump inauguration. [3:56] Mr. Huang made a big donation to Trump's favorite project, [3:59] his gold-encrusted ballroom. [4:02] Mr. Huang wheedled his way into a seat on Air Force One [4:06] on the President's summit with President Xi in China. [4:10] Mr. Huang even managed to score a meeting with President Xi himself, [4:15] along with President Trump, about NVIDIA's business interests. [4:20] All that sucking up to the President has paid off big time for Mr. Huang. [4:26] America had strong export controls that would have blocked NVIDIA [4:31] from selling these chips to China, [4:34] but the Trump administration is relaxing these rules. [4:38] The Trump administration has also turned a blind eye, [4:42] as massive numbers of these chips have been sold elsewhere, [4:46] then somehow magically made it to China. [4:50] The American people deserve to hear Mr. Huang answer hard questions [4:56] from both sides about the role he has played [4:59] in undermining our national security. [5:04] AI holds tremendous promise, [5:07] but Americans are rightly concerned [5:10] that it could further rig our economy. [5:13] Congress can no longer be a bystander, [5:16] as the risks from AI grow nearer, [5:19] and millions more American chips are sent to China. [5:24] It is time for Congress to stand up, [5:26] so that the American people can share in the success of this technology, [5:31] and not just be left holding the bag for whatever goes wrong. [5:35] Thank you, Mr. Chairman. [5:37] I'll now introduce the witnesses quickly. [5:39] Today, we have before us Mike Flynn, Senior Vice President of ITI. [5:43] David Feith, Senior Fellow at the Hudson Institute. [5:49] We have Will Reinhart, who is a Senior Fellow at the American Enterprise Institute. [5:55] We also have Dr. Sarah Myers West, who is the Co-Executive Director of AI Now. [6:02] First, we'll hear from Mr. Flynn, you now have the floor for five minutes. [6:08] Chairman Scott, Ranking Member Warren, distinguished members of the Senate Banking Committee, [6:14] my name is Mike Flynn, and I'm a Senior Vice President at the Information Technology Industry Council, or ITI. [6:21] Thank you for inviting me to testify today. [6:24] ITI represents 80 companies at the center of American innovation, [6:29] including companies that operate at every layer of the AI stack, [6:34] chip makers and designers, leading frontier AI labs and cloud service providers, [6:39] cybersecurity firms, and the software developers that bring AI to consumers in the U.S. and around the world. [6:47] ITI's members are investing hundreds of billions of dollars in the United States to build semiconductor fabs, [6:55] conduct research and development for new AI applications, [6:59] and drive innovation at every layer of the AI stack. [7:04] Artificial intelligence is the most significant driver of economic growth in a generation, [7:09] and the competition within ITI's membership and the broader IT ecosystem is fierce. [7:16] Frontier AI models rely on thousands of highly skilled engineers, advanced semiconductors, [7:22] with features many times smaller than the width of a human hair, [7:26] and networking that operates at the speed of light. [7:29] To name only a few of the vital inputs needed to build and run a frontier AI model, [7:35] we should view the American AI stack as an asset to U.S. economic security. [7:41] Congress must not take this for granted. [7:43] U.S. companies need clarity and certainty to sign multi-million dollar contracts with customers, [7:50] spend billions of dollars on custom chips, and continue the trillion dollar investments [7:55] in America that are making the U.S. the undisputed AI leader. [8:00] The question before this committee is how to protect American AI leadership while promoting adoption [8:08] of the U.S. AI stack around the world. [8:10] The AI supply chain is highly specialized, resource-intensive, and globally complex. [8:18] An AI server in an American data center providing AI to customers in South Carolina depends on memory [8:26] from South Korea, chips from Arizona made with tools designed in Massachusetts. [8:31] A semiconductor may touch as many as 70 countries before it reaches a consumer. [8:37] Key parts of the supply chain are based in allied nations in Europe and Asia, [8:42] especially South Korea, Taiwan, and Japan. [8:45] Because of this reality, the United States must work with like-minded nations on economic security policies [8:52] like export controls. [8:54] The agency Congress has tasked with this important work is the Bureau of Industry and Security, or BIS. [9:00] To meet the demand of this moment, BIS needs appropriate resources and oversight. [9:08] Recently, ITI endorsed six bills that advanced this through the House Foreign Affairs Committee. [9:13] These bills are a down payment to improve BIS's ability to fight diversion of controlled items, [9:20] ensure timely licensing reviews, and develop AI export controls side-by-side with industry experts. [9:27] Effective export control policy and enforcement must be part of a broader strategy, [9:32] that enables the U.S. technology sector to remain agile and competitive in global markets for the long term. [9:39] With the remainder of my time, I'd like to touch on one other issue that I know is top of mind for this committee, [9:46] and that is the cybersecurity of financial institutions. [9:48] Frontier AI models are developing and accelerating the pace and presenting new challenges for cybersecurity professionals. [9:57] That is why it is more important than ever that industry and government work together to address these emerging risks. [10:04] One thing Congress can do right now is to ensure a long-term reauthorization of the Cybersecurity Information Sharing Act, or CISA 15. [10:13] This is a critical law that protects the ability of organizations to share cyber threat information and vulnerability information, [10:21] not only with the government, but with one another. There is more work to be done. [10:26] And we welcome the partnership of this committee on these serious questions of U.S. economic security. [10:32] Thank you, Mr. Chairman, for convening this hearing. I look forward to the committee's questions. [10:36] Thank you, sir. Mr. Feith. [10:41] Chairman Scott, Ranking Member Warren, and committee members, thank you for the opportunity to testify. [10:46] My focus is AI competition with China, and how export controls can help preserve American security and other interests. [10:53] We face a historic opportunity today with three aspects. [10:58] First, the AI industrial revolution will shape not just American civilization, but human history. [11:04] It's hard to overstate the need to get AI policy right, especially regarding our strategic challenge from China. [11:11] Second, we're at a hinge point in Trump administration AI policymaking, [11:15] which rightly recognizes that America must win the AI race with China. [11:20] Recent events involving the super powerful mythos model, [11:23] ship smuggling, and export control loopholes show why we need better strategy, rules, enforcement, and oversight. [11:30] Third, this committee has a particularly key role to play. [11:34] Members here and across Congress have introduced an admirable set of bills [11:38] to strengthen America's export control system, bolster national security, [11:42] and assert Congress's role in this key area of policy. [11:46] What's at stake in the strategic contest over AI? [11:50] One essential feature is the personalization of technological power. [11:54] This is in some ways very positive, in some ways very threatening. [11:58] AI offers personalized medicine, but also personalized warfare, and especially ominously personalized state power. [12:07] Will this technology ultimately advantage systems that protect human freedom or systems that stifle and tyrannize? [12:14] There is some good news to start with. [12:17] In the US-China competition in AI, the decisive factor is likely to be computing power or compute. [12:24] Here, the United States and our allies enjoy a substantial advantage. [12:28] Today, China's best AI chips are much less capable than America's, [12:33] and thanks largely to export controls, their chips are falling further behind. [12:36] Compute is the lifeblood of AI, and China has less of it at lower quality than do the United States and our allies. [12:44] That makes compute China's principal choke point in the AI contest. [12:48] But this advantage is not self-sustaining. [12:51] It will last only if US and allied policy preserves it. [12:55] This is where export controls come in. [12:57] Unfortunately, current US export control policy has not been equal to the task. [13:02] The Trump administration's top-line policy is to prevent America's most advanced chips from going to China. [13:08] President Trump has said this repeatedly. [13:09] His administration's AI action plan also calls for tougher export controls and more allied cooperation. [13:15] But these policies are not yet realized. [13:18] Instead, our stand on exporting advanced AI chips to China has become less clear. [13:23] Commerce Department entity listings haven't kept pace. [13:27] Controls on semiconductor manufacturing equipment, perhaps the most important issue of all, remain loose. [13:33] Remote access remains unaddressed. [13:36] Amid signs of massive smuggling, especially via Southeast Asia, we have seen a sensational federal indictment, [13:41] but little adjustment from our export control system. [13:44] And we have only recently learned that alarming loopholes in US rules were evidently created last year, [13:50] allowing sales of the most advanced US chips to Chinese firms outside of China, [13:54] and maybe also allowing China access to leading-edge fabrication at allied firms like TSMC. [14:01] Finally, on distillation attacks, supercharged IP theft for the AI age, [14:06] we risk repeating 20 years of past failure to counter China's hacking, [14:11] hence the historic opportunity before this committee. [14:15] I'd recommend enacting the set of complementary bills now before the Senate, [14:19] including the AI Overwatch Act on exports to China, the Match Act on equipment, [14:24] the Remote Access Security Act, the CHIP Security Act on location verification, [14:28] the Deterring AI Model Theft Act on distillation, and the CHIP-Equip Act protecting American fabs at home. [14:36] Each addresses a key piece of the puzzle. [14:39] Also, enhance oversight of commerce, draw on the classified record, [14:44] and real evidence from export control enforcement in the field. [14:49] Finally, beyond export controls, you have an opportunity to codify the Commerce Department's ICTS authority, [14:54] and help protect US data center infrastructure from sabotage by Chinese exposed hardware. [15:03] Key technological developments have changed world history. [15:07] Iron, gunpowder, nuclear weapons. [15:10] AI is another. [15:12] The future now turns on whether America, Western civilization, [15:16] and liberal democracy will lead the age of AI, or communism, oppression, and state power will. [15:21] That is why these issues matter so much. [15:25] America should innovate at home, export confidently to allies and trusted partners, [15:30] and deny China the capabilities it needs to overtake us. [15:33] Thank you. I look forward to your questions. [15:35] Thank you, sir. Mr. Reinhardt, the floor is yours. [15:40] Chairman, ranking member, committee members. [15:42] As a kid, I was obsessed with my grandfather's red tractor, a classic Farmall H. [15:47] For decades, you could find it parked in between the garden and the barn, [15:50] basically, you know, always pointed towards the house, [15:53] acting like a sphinx, protecting the cornfields. [15:56] I've since learned that the Farmall was truly revolutionary for the farms and families [16:00] where I grew up in central Illinois. [16:03] Tractors first spread through the wheat belt in the 1920s, [16:07] but it took another two decades for the tech to be adopted for use in the Corn Belt. [16:11] Farmalls were especially important for small to medium-sized family farms because [16:15] they were high-quality but affordable tractors that could do a range of different tasks. [16:20] It took years, however, for those farms to truly adapt to the new technology and to make it useful, [16:25] and to change how farming was done in the Midwest. [16:29] As I see it, AI is in its wheat belt stage. [16:32] It is having an impact in some very narrow and important categories, [16:35] but yet it is yet to be transformative to the broader economy. [16:40] As I've detailed in my written testimony, one of the biggest trends that I'm seeing currently right now [16:44] in AI is that it's a skill equalizer. [16:47] In customer support, software development, legal analysis, writing, and education, [16:52] AI continues to do its best by boosting those who are at the bottom. [16:56] But when tasks fall outside the model's current capabilities, when workers are already at the frontier, [17:01] AI can do more to hurt performance than to help it. [17:06] Importantly, judgment still matters. [17:08] So as I see it, Congress really should be thinking about two things, [17:12] scenario planning and better data sources. [17:15] For this committee in particular, planning around agentic AI could be very useful. [17:19] Agents are only now getting deployed, and I know there's been some discussion of this, [17:22] but they could have some incredible and important impacts in finance and banking and commerce. [17:27] That's something that I have been focused a lot on, and I believe you should as well. [17:31] But we also need better data sources. [17:33] Right now, the data on AI usage only has effectively a 15% response rate. [17:38] There are better ways to collect information, especially through payroll data and other sources [17:42] that are currently being deployed. [17:45] But I would say that it's this, right? [17:47] The Farmall eventually transformed farming across the Corn Belt, but it required the right conditions. [17:52] Congress can help create those conditions for AI. [17:54] The opportunity is very, very real, but so is the work that's required to do it. [17:59] So thank you, and I look forward to your questions. [18:01] You, John? [18:06] And please, on behalf of the chairman, Dr. Sarah Meyers-West, you're welcome with your opening remarks. [18:11] Thank you. [18:14] Chairman Scott, Ranking Member Warren, and esteemed members of the committee, [18:18] thank you for inviting me to testify on this important set of issues. [18:22] My name is Dr. Sarah Meyers-West, and I, along with my colleague Amba Koch, lead the AI Now Institute. [18:27] We're an independent policy research center founded in 2017 to study artificial intelligence technologies. [18:33] I also previously served as a senior advisor on AI at the FTC, where my role focused on the [18:39] effects of AI as they were emanating across the broader economy. [18:43] There's a tremendous amount of hope articulated in the vision for AI that it will have transformative [18:49] potential that could lift our economy and improve the quality of life for all of us. [18:54] We're equally seeing a tremendous amount of pessimism from the public about what this technology means for [19:01] their lives and their livelihoods. This divide is because there are skewed market incentives [19:07] incentives that have shaped corporate AI development and that are driving concentrated risk in the [19:12] economy that the public is being asked to absorb. Importantly, many of these harms are already here, [19:19] and they're being felt by everyday people, and they're going to continue out into the future, [19:23] especially if things go exactly as the industry intends. My message to this committee is that we must put the [19:29] needs of the public back at the center of the AI economy. Congress has an obligation to intervene before [19:36] an economic crisis occurs, starting by alleviating the burden that people are already feeling from AI, [19:43] exercising congressional oversight over AI firms that have grown in size and in political influence, [19:50] and stopping a bailout of the industry that is arguably already underway. There are three core areas [19:57] that I urge this committee to consider as priorities for intervention. First, addressing the risky and [20:05] capital-intensive nature of AI investment. AI companies are engaged in a race to the bottom, [20:11] one in which firms have exhibited an extraordinary willingness to court risk that's not just borne by [20:16] the firms themselves, but by the wider economy and the public. AI development is increasingly being [20:23] financed through complex and opaque instruments that hinge on the private credit market and are often [20:29] kept off balance sheet through the use of special purpose vehicles. We're also seeing circular spending [20:34] deals that hinge on the projected future revenues of just a handful of AI companies, and if those revenues [20:41] fail to transpire to the tune needed to justify this capital expenditure, the house of cards could come [20:46] falling down. Second, the effects of an AI bubble would be acutely felt by the American people. It could wipe out [20:54] a tremendous amount of American household wealth, and this is because the economy at large is already [21:01] significantly exposed to AI risk across several vectors ranging from increased market exposure for [21:07] everyday families, to access to credit, to job loss. We don't have to wait for a financial collapse to see [21:13] these effects of AI play out across the economy. Working class Americans are particularly exposed to an AI [21:19] driven market downturn. For one, the private credit companies have made a casino of American workers [21:26] financial security. We're seeing a transfer of wealth from the public to these AI firms, devaluation and [21:33] dislocation already emerging in the job market, and rising prices, especially in the utility sector, [21:38] that could serve to depress consumer demand. Importantly, many of these factors will remain [21:44] even in a scenario where there's no bubble burst. The vision for success in AI hinges on the significant [21:50] displacement and devaluation of jobs across the economy. I want to underscore if the vision for [21:57] for AI fails, we could face a market correction. If it succeeds, it looks like mass displacement, and this is not a bargain [22:04] any of us need to accept. This is why it's important for Congress to stop the bailout of [22:09] the AI sector that's already in many ways underway. AI firms have received extraordinary subsidies from [22:16] federal backing of loans, to direct funding, to allocating federal lands for data center infrastructure, [22:25] and this is all before Congress has had the chance to meaningfully act to protect the public and the [22:30] economy from the risks posed by the industry. The public's already aware of the risks that they're being [22:36] asked to assume, and this is fueling the generalized antipathy that they're expressing towards AI. [22:42] If the end goal is to promote American innovation and to build an economy that works for everyone, [22:48] Congress must act. It must start by focusing on measures that can alleviate the economic burdens that are [22:55] already being felt by the public, by shedding light on AI industry financing structures through the use of its oversight authority, [23:02] and most importantly by committing to stop the bailout before it deepens any further. Thank you. [23:06] Thank you very much. I appreciate your testimony today. I'll start by asking the first questions. [23:16] Our economy will benefit immensely from American companies' leadership in AI and advanced technology. [23:22] However, we received a letter just recently today from Anthropic that Alibaba executed the largest known [23:30] distillation attack on Anthropic to date. That means that our adversaries are willing to lie, [23:35] cheat, and steal to get access to the information necessary to make China the most dominant player in [23:42] the market as opposed to the free market economy that we have that produces the world's greatest thinkers [23:48] and innovators. I think that's a very important point because when China lies, cheats, and steals, [23:55] the American consumer pays the ultimate price for that. When I'm in restaurants in Charleston and around [24:01] the state, one thing I hear a lot about is the cost of electricity, water usage. There is this fear [24:08] associated with higher prices in an economy where it's very difficult for people to live paycheck to [24:15] paycheck. And so when you put together these amazing attacks from our adversaries, the fear of our [24:22] consumers, it puts us in a position where, as a committee and as responsible leaders, we need to [24:28] have a better appreciation for how the costs can come down and how to push back against the threats [24:35] that could drive our costs up and make us less relevant. Mr. Flynn. Senator, thank you for that [24:42] question. And I think you laid out the depth of the challenges really well. When you think about the [24:47] distillation attacks in particular, that's a real challenge that the extent to which China will go. [24:57] To me, the answer is about where is the US AI stack? Are we seeing it in Europe and the global south? [25:03] And I think it's worth saying that's not a guarantee. The Spanish Ministry of the Interior just recently [25:08] signed a significant deal with Huawei, despite the fact that the US government over several years has done a lot [25:14] of education and advocacy. Going back to the 5G conversation not that long ago about the dangers [25:19] of having the cybersecurity risks of having Huawei in their core infrastructure. Still, it seems like [25:24] we're having that conversation all over again. Mr. Reinhart, ChatGPT is socialized throughout all of our [25:35] devices, it seems like. And of course, all the other ones are working really hard as well. How do we [25:40] help the average family in this country benefit from lower prices, more information? I'll use this as an [25:50] example. Someone was looking for an air conditioning unit at home and they used these models to do [25:58] research. It would take several hours in 90 seconds. I think so we have a lot of opportunities to share [26:06] with the American people, the benefits, the opportunities that come with this technology. [26:11] I'm not sure that the information is getting to the end user, but certainly the benefits should. [26:17] Yeah, the figuring out how to use technologies properly is something that is just a it's a it's [26:23] a thankless task and something that is very difficult to do. I mean, you know, I work with this every [26:27] single day and even I'm, you know, not completely sure of some of the capabilities. But we do see a lot of [26:32] people using, as you mentioned, we see people using agents, as they're called, to figure out the best [26:37] prices for airlines, for a whole bunch of range of different products that they're buying. And to me, [26:42] that I think is a so it's a really prime benefit of these AI technologies is giving consumers choices [26:50] and showing them what's actually happening in the marketplace and giving them the ability to choose [26:55] the better price and things that they work for their families and for their for themselves. [26:59] Excellent. Mr. Feith, what is the single biggest bottleneck to maintaining U.S. leadership in AI [27:09] right now? Thanks, Senator. I think it's it's compute broadly, which includes both the chips and [27:18] the chip making equipment, very crucially. You said that in your testimony. That's one of the reasons [27:22] why I called on you to ask that question. Help me understand perhaps the larger perspective [27:28] on export controls and the role it should play in keeping maintaining America's stronger position, [27:37] especially against China. But perhaps the part of the question I would like for you to answer in [27:42] 25 seconds, by the way, is how do we make sure that our allies don't have a back door, [27:50] whether in the Middle East or Asia, that goes to China and gives them access to the most powerful chips? [27:56] Now you have 17 seconds. It's important. There's a very big allied component, which partly reflects [28:02] the strength that the U.S. and our allies together have in the leadership in these key technology [28:08] areas. That's in semiconductor design and in the manufacturing equipment. But getting allies, [28:14] including the Dutch and the Japanese, especially on board, is really very important. And we have [28:17] only a mixed record of doing it over the last several years. You did what I could not do in 23 seconds. [28:22] Good job. Regan Member. Thank you, Mr. Chairman. So we need to address the serious national security [28:29] risks of allowing advanced AI chips to end up in China. Last week, the New York Times reported that [28:37] NVIDIA's chips have been, quote, sought by nearly every branch of the Chinese military. And China's been [28:46] getting its hands on this technology. Federal prosecutors allege that billions of dollars of [28:52] chips and servers have been unlawfully diverted to China during the Trump administration. Now, [28:59] I invited NVIDIA's CEO here to testify. He has claimed there is no evidence of chip diversion and that the [29:10] Chinese military doesn't want his chips. But he will not show up here to repeat those claims [29:16] under oath. Mr. Fyfe, you served on President Trump's National Security Council until last year. [29:23] How worried should we be about these chips falling into the wrong hands? [29:28] Thanks, Senator. I think very worried. I mean, the implications are very significant in ways that you [29:33] pointed out in part, military, domestic repression, also China's commercial strategic ambitions. [29:40] There's an important additional aspect, which is that if we sell them these chips, [29:43] it actually helps China compete against us in third markets around the world, helps them export [29:48] their cloud, for example, and undermine the market for US cloud exports all around the world when we're [29:54] interested in otherwise maximizing the exports of the AI tech stack of the United States around the world. [29:59] So bad, bad, bad is what I'm hearing you say. Now, Congress has pending legislation to put a stop to [30:05] that. Senator Banks and I have introduced the AI Overwatch Act, which would ban the export of advanced [30:13] AI chips to China. Senator Kim and I have co-sponsored the Match Act, which would ensure China's not getting [30:21] the equipment it needs to manufacture advanced chips on its own. And Senator Cotton and I have co-sponsored [30:29] the chip security act, which would ensure that we can track when AI chips are smuggled. Mr. Feith, [30:37] do you think Congress should pass these three bills? Yes, and kudos on the bipartisanship. Good. [30:43] Mr. Chairman, there's broad support for these bipartisan bills. I think we should mark these up, [30:50] this legislation this month, and I hope you'll have a chance to agree on that. Well, the gentleman from [30:56] Massachusetts yield the rest of your time and all. No. Okay. So I want to turn, though, to how the AI [31:05] boom is being funded. AI companies are on track to spend a whopping $7 trillion on AI infrastructure by 2030 [31:13] and are turning to shadowy financing alternatives to meet their financing needs. If the AI companies can't make [31:21] payments on their huge debt loads, they could trigger massive losses across the financial system. [31:29] Dr. Myers West, if an AI bubble bursts, what would be the impact on families who may not have a single [31:38] dollar of investment in an AI company? Would they be hurt? The impact on them would be significant, [31:44] even if they haven't directly invested in AI, because we could see compounding effects across [31:51] people's retirement funds, their pension funds. We'll see job layoffs and a potential credit freeze. [31:58] The underlying conditions of our economy make us very ill prepared to deal with this kind of market [32:04] exposure, and our social safety net is not well prepared for these kinds of conditions. So it could [32:09] be a perfect storm that will be felt most acutely by people who are already struggling to make things [32:14] work, and it would be irresponsible to look away. Thank you for that answer. You know, AI executives [32:20] are already well aware of what's at stake, and they are right now trying to lay the groundwork [32:27] to leave American taxpayers holding the bag if their big bets go bust. OpenAI, for example, has pushed the [32:36] Trump administration to, quote, lean in and assist the industry by, quote, de-risking AI expansion, [32:46] including expanding federal tax incentives and loan guarantees for AI companies. So I'd like to get [32:53] our panel of experts on the record and just go down the line one sentence or even one word. Would you [32:59] support the Trump administration bailing out big tech CEOs and AI companies? Dr. Meyers-West? [33:09] I would not support a bailout of the AI industry. Mr. Reinhardt? Goodness, no. I like that. Mr. [33:17] Fein. Hard to generalize. I would say there are some funds that come from places like the CHIPS Act [33:22] that the Commerce Department still has and could use very productively. I understand that, but do you want to [33:27] bail out AI if these bets go bad? I think the CHIPS Act on on-sharing semiconductor manufacturing is a [33:33] better bet. Mr. Flynn. I would agree with David. I would also say that I haven't seen any particular [33:40] proposal. I'm sorry? I haven't seen any particular proposal, but happen to take it back. Happy to take [33:45] it back. I'm out of time, but I got to say, trying to backpedal on the question of whether or not these [33:50] guys should make trillion dollar bets and then be able to turn around and say to the American public, [33:56] the taxpayers are going to bail us out. We did that in 2008. We shouldn't be doing it again. [34:01] Sorry, Mr. Chairman. I apologize. Senator Realms. Thank you. First of all, [34:07] let me just thank you all for being here. Second of all, I really appreciated the opening statements [34:11] that all of you provided. Mr. Flynn, I'm just going to begin with you. Do you know of any bailouts [34:18] requested by any of the current AI companies? Not to my knowledge, Senator. Does anybody know of any [34:26] bailouts that are being requested by any of the AI companies? See, here's the deal. I think it's fair [34:34] to say in this country, we recognize that artificial intelligence is here and it's not going away. [34:40] The real question is, is China going to lead in AI development and take the opportunities, [34:47] or is the United States going to take the lead and be the place where the AI development actually occurs? [34:53] In the United States, we believe that it should be for the betterment of everybody and that the [34:57] developments are such that it will grow our economy. Mr. Flynn, has the development of artificial [35:03] intelligence in the United States added to our GDP? Absolutely, Senator. Mr. Feith? Yes, [35:08] sir. Mr. Reinhardt? Indeed it has, yes. Dr. Meyers-West? We've seen a number of places where the [35:17] American public has already been feeling the pain. You can't tell whether or not AI has actually [35:22] developed or improved on GDP growth in the United States? The predominant factor has been the investment [35:29] in AI infrastructures rather than AI itself. You see, what we've got right now is in the United States, [35:34] we've done everything we can on a bipartisan basis to make sure that the United States does not fall [35:39] behind. If we don't do this correctly, AI will be used in the defense of our country, but it'll also [35:46] be used by China in offensive operations. Today, in the United States, we use it in both offensive and [35:53] defensive operations. We're better than our adversaries are right now. That saves American lives long term. [35:59] But if we're going to do that correctly, we have to make sure that this is the place where those [36:03] developments occur. That means that the rest of the world, which is also a market for artificial [36:08] intelligence and the systems that literally people are going to use, they're on the rails that are [36:15] developed in the United States today. We don't want those rails to be China's rails. We don't want Huawei [36:21] doing it. We don't want the companies there doing it and taking advantage of it. So, Mr. Flynn, what happens [36:27] if an organization such as NVIDIA is not allowed or is restricted from being able to market their [36:33] products around the rest of the world? Does that mean then that China then has an advantage in terms [36:37] of selling to the rest of the world? Senator, I think this is a core question [36:41] of this hearing. There's an inherent tension that needs to be addressed. And this is why the role of [36:45] BIS is so profoundly important, not just on the export control side, on the enforcement side, [36:50] but their ability to issue licenses in a timely fashion and to work with the regulated community, [36:55] so the information necessary to ensure that the U.S. technology stack is being used around the world. [37:01] Mr. Feith, I really appreciated your background and the fact that you've indicated that while we need [37:07] to continue to lead the world, there also needs to be the appropriate way to make sure that we keep [37:13] that away from other adversary companies such as China. Today, would it be appropriate if we were to [37:20] get the Commerce Department additional resources to be able to monitor and to restrict that from getting [37:26] into the wrong hands in China? Yes, certainly. Okay, let me let me run this by you again once on this [37:32] thing here. Is there any reason that we should be trying to chase any one of these organizations, [37:37] these hyperscalers, any reason why we should chase them out of the United States by trying to tax them [37:42] at a higher rate than other businesses would be taxed? No, I don't believe so. Mr. Flynn? No, Senator. [37:48] Is there any advantage at all to the American people if another country, including China, takes the lead [37:54] on the development of artificial intelligence in this once in a generation opportunity to advance [38:00] GDP in our country? Certainly not. Goodness, no, I do not. I fear a world where China is leading on AI. [38:10] Can you imagine a world in which if you use artificial intelligence correctly, you can actually improve [38:15] the quality of life for Americans? We don't talk about it much, but what happens as we integrate [38:21] biomedical research and the development of artificial intelligence, being able to make decisions at a very, [38:26] very rapid pace. Mr. Flynn, does that make a difference for the quality of life for the [38:30] American families? Absolutely, Senator. Mr. Feit, do you agree with that? Yes, sir. [38:35] Mr. Reinhardt, you've been kind of quiet today, but would it be fair to say that there are real [38:40] opportunities for the American public to benefit from the development of artificial intelligence in [38:45] the United States rather than in China? Yes, very clearly, and I would say one thing I saw recently, [38:51] an individual had a very rare complication to a heart medication, and by putting it into CHAP GPT, [38:58] he realized that he needed to switch his heart medication. So I think that these small stories [39:02] are actually incredibly important for individual lives. Thank you. Look, I'm running out of time, [39:06] but here's the deal. Artificial intelligence is here. It's not going away. We've got to be able to do the [39:13] right types of guidelines for the development of it, but we have to do everything we can to keep that [39:19] development here in the United States, both for economic development, but for the quality of life [39:24] that it proposes for the people that live in this country. Thank you, Mr. Chairman. Thank you, sir. [39:29] Senator Reid. Thank you very much, Mr. Chairman. Thank you to all the witnesses. Mr. Feit, there is concern. [39:38] I heard a bit of Senator Roundt's testimony about providing China with H20 and H200 chips from NVIDIA, [39:51] and one reason, and this is what I've heard from national security experts, is China is very rapidly [39:59] and very adeptly using those chips to expand its very sophisticated and indeed weaponized chips. Is that [40:08] accurate? Yes, sir. I think we've seen valuable testimony from the US military and elsewhere on [40:15] that. Now, they are, China is promoting, and promoting is an understatement, the Huawei chips and [40:24] other chips. They are not, I don't think, bringing the H20 and H200 chips in to displace their own chips. [40:34] On the contrary, I think they're doing all they can to promote their own Huawei chips and others. [40:39] Is that right? That's right. They want very much to indigenize, [40:43] and they want to gain so that they can not only satisfy their domestic market with Huawei chips, [40:49] which right now they're way behind in their ability to do, but their ambition is to do so, [40:53] and then beyond that they want to then compete to export all around the world [40:57] against US chip makers and cloud companies. So is it fair to say that what they really want from [41:03] H20 and H200 chips is to get the most advanced American technology they can to incorporate in [41:10] their chips and also to incorporate in their sophisticated weapon systems? [41:16] Yes, and I would say that there are other chips, including the Blackwells, that they've demonstrated [41:22] a major interest in getting as well, even though they're not licensed for export, like through smuggling. [41:26] So the president's decision to allow them to get H20 and H200 chips, what would you consider the rationale [41:39] to be if in fact they're using it to exploit our technology, not for any other reasons perhaps? [41:48] Well, I think the administration stated that their calculation was that these are not the leading [41:56] edge chips, they are second or third level chips, and that sending them to China would therefore not [42:04] give China the best, but could complicate China's ambitions to indigenize. My sense was that the [42:11] calculations probably cut the other way, but I do think it's important that the administration has been [42:15] very clear. The president has said repeatedly that the principle is to deny China the leading edge [42:20] chips, to deny them Blackwells. That's been maintained. I think it's important to follow it up [42:26] with robust enforcement, including around smuggling and on possible loopholes. [42:31] No, I think that makes sense, but I think as you point out, the H20 chips, et cetera, are not going to be [42:41] indigenous to China. They don't want American chips to be dominant in their market, and they [42:47] are exploiting these chips, and they may not be the most exquisite chips we have, but frankly, [42:53] they appear to be better than what the Chinese have, and they're weaponizing them. Is that fair? [42:58] That's right, and I think part of your question also points to this question of, you know, [43:04] what is China really about? Is it about simply maximizing their compute access and potentially [43:11] maximizing business for U.S. exporters, or is it about indigenizing and replacing? And I think China's [43:16] ambitions in semiconductors are as they have been in other industries. They will only take U.S. [43:21] technology as long as it takes them to indigenize and then overtake us. We shouldn't think that we [43:28] can addict them to these U.S. exports. Thank you very much, Dr. West. Dr. West, [43:34] we have to think seriously about the impacts, particularly in employment impacts, of the wide-scale [43:42] introduction of AI. I know the former Secretary of Commerce, Senator, Congress Secretary, excuse me, [43:49] Raimondo wrote a very good editorial in the New York Times indicating we're not ready yet for the [43:57] impact. Could you comment on that? Yes, I mean, I think we're already seeing, you know, firms narrating [44:05] layoffs as due to AI, whether or not the technology's in fact the causal factor. And here, I think it's critical [44:12] that workers who are sort of on the front lines of experiencing the impact of AI, whether that's [44:18] being laid off or their jobs being devalued and their wages depressed, be positioned to have a [44:24] crucial say in the terms on which AI is deployed in their workplaces. And thank you very much. And I'll [44:31] yield back my second. Thank you. I'll give it to the ranking mover for five seconds. I just want to say [44:39] that when I was talking through the bills, I failed to mention that the lead sponsor on match is Senator [44:44] Ricketts. And it was just an oversight. I apologize. It's a terrific bill. Senator Kennedy, the floor is [44:51] yours, sir. Mr. Feith, is that how you say your name? Yes, sir. I continue to read and hear that we have to win [45:10] the AI race. What does that even mean? Generally, we heard that with 5G. Generally, we hear that when [45:27] somebody in the private sector says, we're behind China, give us some money. If we are ahead, [45:36] eventually China's going to catch up. And then China gets ahead, and we catch up with them. Ultimately, [45:48] and there's no finish line here where the first one who crosses gets a gold medal. This stuff keeps [45:58] evolving. Aren't we ultimately going to have to sit down with China and reach some sort of compromise [46:10] in terms of a modicum of cooperation? Well, I think, Senator, I think the questions about whether this race [46:20] vocabulary and metaphor are exactly right, are very well taken. I think it's inexact at best. But [46:32] there are comparisons possibly with previous generations of strategic technology like nuclear [46:40] weapons, where I think maintaining that the US and our allies have a stronger foundation of strategic [46:47] technology, including as it needs to be innovated across generations and including as, you know, after the [46:53] Soviets got a nuclear weapon in their time several years after us, is extremely important. And the [46:58] fact that this is commercially innovated... I got it. I'm sorry. I don't mean to be rude, but I don't [47:03] have time for history. Dr. Winston, I want to make sure you're heard. Well, before I come to you, [47:17] let me ask Mr. Reinhardt, do you think we ought to have state-owned AI companies? Are you a fan of the [47:28] state-owned companies in China? No, not at all. Whether the federal government takes a big stake [47:35] and runs the show? No, I do not. We can't even deliver the mail, can we? Especially in my [47:42] neighborhood, yeah. And the stuff has an address right on the front, doesn't it? Yeah, I think that [47:49] we can be smart about how we do grant programs. I mean, there's ways that we can be smart about the... [47:53] Grant programs? Well, just the various, you know, we've had, as we've talked about, some of the CHIPS [47:59] Act, parts of the CHIPS Act program are still out. There are... I understand there are alternatives [48:04] to grants. They're called loans. Of course. Of course, yes. Okay. Yeah. I'm just saying that I [48:09] think there's a lot of, there's lots of parts of the federal government that we can think about being [48:13] smart about our investments in AI, and that doesn't necessarily require any sort of ownership. Well, [48:19] look, I think the people developing AI, this is true worldwide, but it's certainly true in the [48:27] United States. These are brilliant people. Right now, they're spending money to try to be the biggest [48:42] and the best and the wealthiest. And I don't mean disrespect when I say this. I'm proud of them [48:48] as American entrepreneurs, but I trust them like I trust the rest stop bathroom. I remember when [48:57] when one of the leaders in another hearing about a year ago, I asked him, I said, man, you, you know, [49:04] you must make a lot of money. And he said, oh, no, no, no. I do this for the goodness of the people. [49:11] I only get my health insurance paid. Wrong. So, Dr. West, you make a good point, but I guess my question is, [49:23] what do you expect us to do? It's here. And what do you, people are going to lose their jobs. [49:30] I hate to see that. I might lose mine. But what are we supposed to do about that? [49:40] I think it's a great question because there's this sort of generalized environment that this is [49:44] inevitable and we all have to just kind of deal with the consequences. [49:47] Yes, ma'am. I understand the problem. I'm looking for, I admire the problem. I'm looking for a solution. [49:53] Tell me what we think we ought to do when somebody loses their job as a result of AI. [49:58] In the next 10 seconds or so. Oh, wow. [50:01] I think we need, we need strong accountability measures that restrain the, the, you know, [50:06] the actions of this, this industry that are sort of experimenting in the wild with the release. [50:11] Dr. West, that tells me nothing. I'm sorry. I'm way over, but you're a good man for letting me go over it. [50:19] Yes, sir. Well, I just wanted to associate myself with your comments about the rest stop bathrooms. [50:28] There's a, there's not a lot of confidence in those. You gotta, you gotta be careful. The coffee's lousy too. [50:32] Yes, sir. Well, hadn't had that experience. [50:35] Especially in Maine, evidently. [50:38] Senator Warner. [50:39] Well, I'd like to invite my colleague from South Carolina and Louisiana. A certain former governor [50:45] of Virginia made it a top priority to improve the quality of all the rest stop bathrooms in the [50:50] Commonwealth. And we welcome you to use our facilities at your discretion. Let me, let me answer. [50:59] I think, you know, I think what Senator Kennedy asked is a really good question. And I've talked [51:04] to every one of the big heads. I spend more time on AI probably next to Mike Rounds of anybody here. [51:09] And there is no macro answer. Four years ago, we thought, well, let's make everybody learn to code. [51:14] That was well-intentioned, but not the right answer. Frankly, one of the things we ought to do at a macro [51:20] level is in high school, junior and senior year, start convincing people that if you're going to major [51:26] in business, business administration, economics, which is the biggest majors in public universities, [51:31] and those are the jobs that are going to be least, most disappearing, we ought to convince a heck [51:35] of a lot more people to go into healthcare, where there will be, there will be jobs for some time to come. [51:42] I'd also say, I think the analogy on the race is a little wrong, too. And I, I absolutely agree with Mike. [51:48] We've got to stay ahead on the, on the leading frontier models. But the fact is, even right now with us ahead, [51:55] the Chinese open source models are 10x cheaper, use a lot less tokens, a lot less electricity. [52:03] And while we're leading most of our smaller entities are already using Chinese models. [52:10] And I hope the panel has seen, there was a Booz Allen study out recently. I think it's quite good. [52:15] And I recommend it to all my colleagues that if you ask the Chinese models a question in English [52:22] or French, anything other than Mandarin, you get a different answer. [52:26] And if you happen to note that I'm working on a government project, you get an answer that comes [52:34] with flaws built in. So the idea that China, we, you know, having been in, in the telecom business [52:40] and seen this movie with Huawei, this is nothing compared to what's happening in Chinese models. [52:46] Where I disagree with some of my colleagues, the idea that we're going to just do pedal to the metal [52:50] regardless of outcomes. I think most of us agree we screwed up by not putting some guardrails on social media. [52:57] This is peanuts compared to where. So the way I'm looking at this, and I apologize, [53:01] I will try to get to a question is, I think we need to realize, I don't think moratoriums on data [53:05] centers make sense. But I sure as heck think making sure there's no misuse of power, water, backup [53:12] generators, setbacks, non-disclosure agreements. So communities, what they're getting into is an enormous [53:18] amount of sense. And they have to pay part of the share. And there's a bunch of ideas around how we, [53:23] how we address that. I'm, I'm looking at an idea that doesn't raise their taxes necessarily, [53:28] but since the benefit will be deferred, maybe we ought to defer some of the accelerated depreciation [53:33] for a couple of years. So we can gain some revenue to help on the retraining as we decide [53:38] what to retrain. We think about social media, social media to your kids is nothing compared to what [53:44] a chatbot can do your teenage son. Where I think back four years ago, we had every one of the AI companies [53:50] promised in elections, we will never do deep fakes, bull, they're out there, they could correct [53:56] that we do need to have them put pedal to the metal. And mythos, thank God it was, it was anthropic. [54:04] When the head of the NSA and cyber commands came and said, this tool broke into almost all of our [54:11] classified systems, not in weeks, but in hours. And one of the things I don't know how we do, and I know [54:18] there's a big debate in the administration, but we're not going to get this fixed if we have a less [54:24] ethical CEO in terms of simply voluntary pre-testing. Now, how we get to the right incentives, [54:31] it doesn't have to be 90 days, it could be 30 days. And I do think the, the dislocation, I'll make a bet [54:38] with any of you here. John, we're at 9% recent college graduate unemployment. I will bet you dollars [54:45] of donuts right now, it'll go to 30% by 2028. And that will have a double whammy, not just the age [54:54] cohort, but you're gonna have a lot of pissed off parents and grandparents who just spent $250,000 [54:58] on an education. And we hear a lot of talk about, you know, AI adjacent jobs, I do believe they will [55:06] come. But we need a whatever, the Democrats will call it a New Deal program, maybe we can call it a [55:12] Reagan era program, something that says for five years how we get through this gap. And we got to have [55:17] the industry help us. If it's simply a government training program, we'll screw it up. So Mike, [55:23] I was going to ask you on CISA and the ISACs, which I think need to be refunded. Mr. Flynn, [55:29] I think you've even mentioned that. And by the way, you've also supported the bill that Senator [55:33] Rounds and I have. Thank you for that. Quick comment on the ISACs and then I'll get out of your hair. [55:36] Profoundly important, Senator. And thank you for so many years of excellent work on cybersecurity, [55:41] elevating that issue set, which is wonky and not everybody wants to talk about it. The ISACs are a huge [55:47] piece of this equation, making sure that the highest leveraged cybersecurity players, [55:50] their information on vulnerabilities of the threat information gets down to players with fewer [55:55] resources is critical to making sure that we're prepared. When I get that response, I'm going to [55:58] leave. Thank you all. Apologize not getting any of the others. Thank you. Senator Ricketts. [56:02] Just want to remind my colleague from Virginia that betting is no longer allowed. [56:05] Senator Ricketts. Thank you. Thank you to all of our panelists for being here today. Communist [56:15] China is trying to dominate AI. They're trying to do it for a variety of reasons. But one of the [56:22] biggest reasons is they think this is how they can dominate us militarily, whether it's through [56:28] dorm swarms, hypersonic missiles, attacking us through cybersecurity, that sort of thing. And [56:36] semiconductor manufacturing equipment, or SMEs, is the ultimate choke point in this competition. [56:44] Many people are familiar with the amazing lithography machines EUV and DUV for printing features on [56:51] chips. Yet SMEs is much more than that. It also includes hundreds of other advanced technologies [57:00] like testing machines, etch tools, chemicals, and specialized design software. Chips models and [57:07] applications all flow downstream from this hardware. Despite pouring billions of dollars into domestic SME [57:14] initiatives, Beijing still hasn't indigenized these highly complex technologies. That's why, [57:20] if you are going to control one part of the semiconductor supply chain, it should be the SMEs. [57:26] Mr. Feith, do you agree that export controls on SMEs are the most effective way to leverage limits [57:31] of the PLA on its AI capabilities? Yes, Senator. Thanks for your leadership on that. The Trump administration [57:37] pioneered using export controls on SMEs to blunt China's chip ambitions. Key allies eventually [57:43] followed. But our lead has dangerous gaps, and those still remain. The controls are narrow and less [57:50] restrictive, and they have loopholes. Foreign companies are backfilling the Chinese market with [57:54] critical dual-use tools. This harms American workers, risks our compute lead, and fundamentally [58:00] undermines our collective national security. As has been mentioned before, and thank the ranking [58:05] member for mentioning it as well, I introduced the Match Act with Senator Kim. Not only in addition, [58:11] I'll throw out Senator Risch and Senator Schumer also co-sponsors on that bill as well. And I would [58:16] encourage this committee, as the ranking member did, to do a markup on this bill so that we can move on [58:21] this as quickly as possible. Our bill prioritizes diplomacy first, giving the state and commerce departments [58:27] a clear window to negotiate with allied governments and get the alignment. But it also establishes a firm [58:33] deadline. If our partners refuse to align, the U.S. must act to stop allied firms from weaponizing this [58:42] technology and giving it to our adversaries like Beijing. Mr. Feith, do you agree that weaker allied [58:49] export and servicing standards on choke points for SMEs create an unacceptable vulnerability in our [58:54] strategy to contain China's AI ambitions? Absolutely. As you said, the Match Act looks to build on [59:00] the first Trump administration's export controls on EUV lithography, which have been by far the most [59:07] effective and crucial across years and demonstrate the enormous benefit that we can get from these [59:12] export controls. And I think expanding that in the way that the legislation points to, to make sure [59:16] that we are aligned with allies as much as possible and that we have additional lithography technologies [59:22] like DUV immersion covered is really very important. The Match Act provides a clear timeline for [59:27] diplomacy, but we cannot negotiate indefinitely while communist China continues to advance. If our [59:32] allies failed to match our prohibitions within the bill's time frame, do you support applying [59:37] foreign direct product rule to close foreign compliance gaps, Mr. Feith? Yes, and I think that [59:43] reflects the record. I mean, it was the ability to do that and to negotiate with that possibility that [59:48] has gotten us the best cooperation we've had over the last eight or so years with the Dutch. We do want to [59:54] negotiate, but we also have to have a hammer, right? Certainly. If we don't act now to close these [59:59] loopholes, we run the risk that Beijing indigenizes and dominates SME just like they did for EVs and [1:00:04] other critical minerals. Do you agree that it is critical that export control loopholes are closed [1:00:10] as soon as possible? Yes, sir. You know, the reason why we should get the markup on this. China's civil [1:00:16] military fusion strategy ensures commercial breakthroughs immediately advance the PLA, [1:00:22] and it eliminates the boundary between civilian technology and military end uses. By law, [1:00:28] if Beijing asks companies must hand over any technology to the military, the Pentagon understands [1:00:34] this is a massive challenge. It just expanded a list of Chinese military companies, the 1260H list, [1:00:42] and added several prominent consumer technology titans. The list restored membership makers CXMT and YMTC, [1:00:52] but it also added three of China's top AI builders, Alibaba, Baidu, and Tencent. Because of this civil [1:01:00] military confusion, it is impossible to track or control any technology once it enters technology. [1:01:08] That's why we have some of these other things like the CHIP Security Act to mandate location [1:01:12] verification technology is a key part of managing this challenge. We need to make sure that we have this. [1:01:18] Mr. Feith, if China's military civil fusion economy, do you support the Match Act's appropriate [1:01:26] country-wide controls as a choke point on SMEs and its efforts to apply these controls to other [1:01:32] affiliates and subsidiaries? Yes, certainly, Senator. And I'll just say quickly, on the 1260H list and [1:01:36] Alibaba, it speaks to the question that Chairman Scott also led with, with respect to Alibaba and the [1:01:42] anthropic distillation, having them identified as military-affiliated companies given military [1:01:48] civil fusion is very important. And we've got export control rules that come into play with that [1:01:54] designation that are also valuable. Great. Thank you. Senator Van Hollen. [1:02:00] Thank you, Mr. Chairman. Thanks for holding this hearing. Thank all of you for your testimony today. [1:02:07] We're all trying to get our arms around a very rapidly moving and developing AI universe with [1:02:14] increasing capabilities. We recognize that these capabilities can bring benefits to Americans in [1:02:21] areas like medical research. And we have embraced innovation and we should continue to do so, but we [1:02:29] should make sure we have the appropriate guardrails in place, in my view, as we do that. Millions of [1:02:35] Americans are asking themselves what this means for their jobs and their ability to earn a living through [1:02:42] hard work. There are questions about whether or not AI will replace humans in making critical military [1:02:49] decisions. I think that would be a mistake. We also have AI agents. And are they going to replace [1:02:59] individual decision making in many areas? So lots of questions. I'm glad we're having this hearing. [1:03:05] One area, of course, is the question of worker displacement. That's an area of concern. [1:03:11] I've been looking at legislation and proposed in the past before the questions of AI arose with respect to [1:03:17] how we can deal with chronic long-term unemployment. And I worry that that could become an even more [1:03:24] acute problem going forward. So Dr. West, if you could just speak briefly to what you see as the risks of [1:03:33] workers in terms of job displacement, a lot of different theories and warnings. But where do you [1:03:41] think we are right now? And would it be useful to begin now to look at how we can prevent chronic [1:03:48] long-term unemployment that is at least part and part caused by AI? I think that's exactly right. [1:03:56] We are already seeing indicators that workers are feeling the experience of job displacement, [1:04:04] whether or not AI is at the root cause. AI is sort of being attributed as the underlying factor. [1:04:09] And our social safety net is just not well set up right now to be able to kind of catch people [1:04:16] when they experience that displacement. And so I think focusing there first and making sure that we [1:04:22] design our unemployment, our other health care, like the other elements of our social safety net that [1:04:32] will help us contend with the potential future impacts on the workforce is an important place to start. [1:04:38] No, I appreciate that. And I think it's, as you say, it's not too early to really give that some serious [1:04:44] thought. I want to turn briefly to a question that I heard Senator Warren raise in her opening statements, [1:04:49] and Senator Ricketts just alluded to it, as others have, which is the issue of our export controls on [1:04:56] very high-end AI technologies and chips. I think we would all agree that AI can be significant in [1:05:07] adding to military capabilities. Is there anybody who disagrees with that statement? And so I have been [1:05:13] concerned about the loosening of export restrictions to some of our chief military adversaries, like [1:05:23] China. There was just a New York Times story the other day that reported that the Chinese military [1:05:28] has sought out high-end chips from NVIDIA. And President Trump has seemed eager to compromise [1:05:35] national security to grant favors to friends and cronies. NVIDIA CEO, Jensen Huang, [1:05:43] wants to sell more chips to China. Like any company, he wants to sell more for his shareholders. And he [1:05:49] has been cozying up to President Trump. He gave money for Trump's ballroom project. He went to a [1:05:54] million-dollar-a-plate dinner at Mar-a-Lago and on and on. And sure enough, President Trump announced [1:06:02] licenses for NVIDIA to sell chips to China, the H20 and the advanced H200 series. And again, [1:06:09] as we've all acknowledged, these have military applications. So we need to be careful. Dr. West, [1:06:17] what can we do? And others may want to answer as well. What can we do to prevent this continued [1:06:23] diffusion of high-end chips and AI technologies that could come back to bite us when they're deployed by [1:06:32] other militaries? I think that you're right to point out the chip problem. I think what I would [1:06:38] add to it is, in fact, that there is a broader challenge faced by the sort of focus on redeploying [1:06:46] commercial off-the-shelf AI models in national security contexts. We've seen the ways that they [1:06:52] have tremendous amounts of security flaws and vulnerabilities. And I think it's important that we [1:06:58] also be addressing the potential integration of those vulnerabilities into the critical infrastructures [1:07:03] in which these systems are being deployed. I really worry about the approach of, you know, [1:07:08] using only voluntary measures to test and validate and look for these vulnerabilities in these systems. [1:07:15] Appreciate it. Anybody else want to weigh in on this? Next. Senator, I think it's a fantastic question. [1:07:20] BIS, strengthening BIS. Export controls are only going to be effective if the Bureau, if the agency [1:07:28] charged with doing that work can effectively administer those programs. And right now, I think [1:07:33] we've seen, at least what I've heard from ITI companies, is a real challenge in BIS's ability to [1:07:39] administer the export controls that are on the books. Thank you. I'm glad you raised that. Thank you. [1:07:44] Senator Moreno. Thank you so much, Mr. Chairman. Thank you for having this hearing. [1:07:49] Just a quick lightning round of questions. Do you think the time has come and gone for data centers to [1:07:54] receive any type of government incentive? I'll lay that up for you, Dr. West. And I assume no. [1:08:01] I think that they've received enough incentive. Right. So no more. So we could just say no more, [1:08:05] right? We could just say any city, county, state does not need to give any form of tax incentive to [1:08:12] any company building a data center. Yes or no? I think these need to be responsive to the local [1:08:17] communities in which they're being deployed. And what we're hearing for those from those communities is no. [1:08:21] So no more money. Mr. Reinhardt? Generally speaking, I'm not a big fan of tax carve-outs, [1:08:27] you know, broad base and low rates. But I do think- Just pay like anybody else. Yeah. Yeah. [1:08:31] Mr. Reinhardt? Sounds right. The average American intersects, [1:08:39] interacts with the data center more than a hundred times a day. Totally understand the conversation and [1:08:45] the challenges that constituents may face in this. And I think that is incumbent on ITI member companies [1:08:53] and our industry to do a better job in educating local populations. Data centers support 5.5 million [1:08:58] jobs in America in 2024. So there are benefits to local communities and it's on us to make sure [1:09:03] that those communities know that. Oh, that's great. So do they get tax incentives? Yes or no? [1:09:07] I'm sure the answer is yes. But they shouldn't, correct? You would agree that that time is gone. [1:09:14] We should not be giving handouts to companies that are building data centers. Would you agree? [1:09:18] Not necessarily. Data centers are an incredible part of- How about car dealers? Should they get tax [1:09:23] tax incentives? The center- They employ thousands of people too. [1:09:28] Absolutely. But my point being that this is, I think, where the anger comes from. [1:09:32] Why would a company that's a multi-trillion dollar company need any money? There was a situation in [1:09:37] Ohio, Carlisle Group came in, built a data center in Akron, and we gave them four and a half million [1:09:42] dollars. Like why? And I've said to David Rubinstein and Carlisle, give them money back. You don't need [1:09:48] the money. And if you need the money, you have a shitty business model. That's the problem. So number two, [1:09:53] I'll start with this crappy business model. See, the best part of being Catholic, Sunday, confession, [1:10:01] it's all over. So switching to power, President Trump put in place the Ratepayer Protection Pledge. [1:10:09] It's very simple. I live in Avon, Ohio. I don't want to compete for electricity with a giant data center. [1:10:16] So data centers should not add to the cost of electricity for average working Americans. Would you agree? [1:10:21] I would agree. Go down. Yes. Yes. Of course. Yes. Yes. [1:10:27] Right. So you represent the industry. So you can, with full confidence, say there's no scenario on [1:10:33] earth in which a data center is built that doesn't either generate its own power or does not have a [1:10:40] increased impact on electricity prices. You would advocate to say you should not do that. [1:10:46] Senator, speaking in general, as I can't speak for every data center in the country, [1:10:50] but certainly ITI is supportive of more electrons than the grid bring your own generation, [1:10:54] getting data centers the power they need. But no impact on electricity. That's really important. [1:11:00] And I applaud President Trump for that Ratepayer Protection Pledge. Number three, environmental impact. [1:11:05] These systems need to be closed loop and have no impact on the environment or a positive impact. [1:11:10] Agreed? Agree. Yes. Yes. Very much so. Sounds right. As a generality. Right. Because we screwed that up in [1:11:20] Ohio. The Ohio EPA put out a statement that said we were going to lower the water runoff [1:11:25] requirements for data centers. That's next level insanity. The last one, when these data centers [1:11:31] come in, they should have a positive impact in the community in some way. What that means is up to that [1:11:37] local community, whether it's workforce grants, improving infrastructure, but they need to be good [1:11:42] public citizens. Would you agree with that? I would agree with that. And I think that the [1:11:46] community should be the leader. Yeah, absolutely. Whatever that community needs. Yes, obviously. And [1:11:52] these data centers do bring in lots of money in property taxes. So I think that there is. But [1:11:56] so does every single solitary commercial business. Of course. Yes, of course. Right. Yes, sir. Yes, [1:12:02] sir. Okay. So these are, I think if we can get through these four principles and data centers behave [1:12:07] this way, the operators, the hyperscalers behave this way, I think we could get this conversation in a [1:12:11] different place. And I think this is very, very common sense. Real quickly, Mr. Feif, I'll ask you the question, [1:12:19] shifting gears completely. We have vehicles that are made by China that connect into critical [1:12:26] infrastructure, record videos, send back data, send back high definition photos. On a scale of [1:12:34] really, really, really crazy stupid to common sense, where would it be to allow these cars to come into [1:12:40] the United States? Full on crazy stupid. Catastrophically stupid. Okay, great. Thank you. You're done. [1:12:50] You gave time back. Fantastic. All right, Senator Smith. Well, thank you, Mr. Chair, [1:12:58] Ranking Member. Welcome everybody to the committee. Dr. West, I'd like to start with you. I'm very [1:13:04] interested in the intersection between AI and housing and fair housing in particular. And I want [1:13:09] to talk with you about that for a little bit. So, you know, Congress passed the Fair Housing Act, [1:13:12] and I believe 1967 or 68. Actually, Walter Mondale, my home state senator, led the way on that. And it's been [1:13:20] a longstanding tenant of our country that we should not be discriminating when it comes to housing. And [1:13:25] so, interesting how AI intersects with this, right? Because we know that AI tools can, you know, there's [1:13:32] ways that they could be beneficial to housing. They could make it easier to build. They could make it [1:13:36] more affordable. They could make it more accessible. But also, experts have warned that without anti- [1:13:42] discrimination protections, they could also, you know, create risks for renters and homeowners. We've all heard [1:13:48] about this sort of, you know, the algorithmic pricing of rent as just one example of that. [1:13:53] So this is just an open-ended question to you. Would you, you know, could you talk a little bit [1:13:58] about this and what you see is the opportunities as well as the challenges of AI when it comes to [1:14:03] housing and particularly fighting housing discrimination? Yeah, I'm so glad that you brought [1:14:10] this into the conversation because it's one of the things that the public sort of directly is [1:14:14] interfacing with AI systems, but the kind of AI that is more used on us rather than by us, right? [1:14:21] It's AI that's used to set the price of rent or to determine what the cost of somebody's mortgage [1:14:25] is going to be. And in those uses, we see a few particular effects. One is that these systems have [1:14:34] a tendency to sort of scale and amplify existing patterns of inequality across society. But in [1:14:42] doing that, they also make it harder for the average American to see the effects. It's harder to see [1:14:48] that you are being discriminated against and to be able to take action. It's sort of like you don't [1:14:53] know what you don't know scenario when it comes to what rent you're being charged and what your mortgage [1:14:58] rate is, for example. Exactly. And it ratchets up the information asymmetries between everyday [1:15:04] people and people who are in positions of relative power. Landlords, you know, mortgage lenders sort [1:15:11] of have greater insight and information and access to data that they can use in ways that, you know, [1:15:19] intentionally or unintentionally have the effect of ratcheting up inequality that everyday people are [1:15:25] feeling. And so we have laws that say that you can't discriminate against somebody when it comes to [1:15:31] housing. But if we're not enforcing those laws, which I think some of us have real concerns about [1:15:35] with this administration, then doesn't that then amplify and amplify the discriminatory impact of [1:15:42] some of these AI tools and also just make it so much harder for people to get any, you know, relief? [1:15:49] I think that's exactly right. It's both the, you know, lack of enforcement compounds with the inability [1:15:57] of individuals to then go and seek remedy. So this is something I think that we need to really [1:16:02] pay attention to, Mr. Chair, as we think about AI impacts. I want to just touch on one other thing. [1:16:07] There have been some very interesting proposals that have come out in the last weeks, I'll say, [1:16:12] around how AI systems that are trained on the collective, you know, knowledge and data of Americans, [1:16:25] that the wealth that is generated by those systems should be shared, whether it's in a sovereign wealth [1:16:30] fund. Senator Sanders has a proposal. I believe even Sam Altman has had a proposal. And so could you [1:16:36] just talk briefly about what you think some of these options are and how we ought to think about as [1:16:42] policymakers how people that are generating the data and the information that is fueling this incredible [1:16:49] boom in wealth for very few people could actually be shared by everybody? Sure. I mean, it's good to see [1:16:58] an interest in approaches that cut right to the heart of AI companies' political and economic power. I [1:17:04] think that that's the right place to be starting from. I think it's equally important that we get the [1:17:09] incentive structures right. And so I think that there's two components here. One is that we make sure that [1:17:16] there are there's an approach and policy that provides for the kind of safety net needed. That's to [1:17:22] you know, help the public to whether whatever changes might be might be coming. But equally that we [1:17:28] also ensure that our regulators have the right kind of accountability measures in place to make sure [1:17:34] that the actual use of these tools isn't then going to cause other kinds of harm. Right. So just to be [1:17:40] blunt about it, if you're creating an incentive structure where the regulatory, the public sector regulatory [1:17:45] function is also, on the other hand, benefiting from the wealth that's being generated by the [1:17:50] entity being regulated, that could potentially be a negative incentive. I think that's the tension that [1:17:55] we need to resolve in the policy approach here. Yeah. Okay. Thank you very much. Thank you, Senator [1:18:02] Banks. Thank you, Mr. Chairman, for first of all, for holding this hearing. There is no contest that matters [1:18:08] more to America's future than the race to beat China in artificial intelligence. And I share your [1:18:13] commitment to this issue. Really appreciate you organizing this hearing. The nation that leads [1:18:18] in AI will set the terms of the world's economy and command the high ground in military power. That's [1:18:25] why I believe that the United States must lead and so does President Trump. America has to make the best [1:18:31] chips, develop the best technology and protect what we build. Our most advanced AI chips are the crown [1:18:39] jewels of American power, and we cannot let them end up in the hands of the Chinese Communist Party. [1:18:46] Our most cutting edge AI chips have been smuggled to Chinese military companies through third nations, [1:18:52] and China is using these chips to build AI models that rival ours. Strong export controls will keep America [1:19:00] ahead instead of handing the lead to Beijing. I have worked on legislation, including the AI Overwatch Act, [1:19:09] with Congressman Brian Mass, the chairman of the House Foreign Affairs Committee, [1:19:13] to make sure that the Chinese military cannot access any of our highest end chips, [1:19:18] just as President Trump has said. Mr. Feith, my bill, the AI Overwatch Act, [1:19:23] is designed to put America's export control enforcement on a firm footing. It would stop the [1:19:29] most capable chips from going to Beijing and building the Chinese military. When President Trump sits across [1:19:34] from Xi Jinping, it strengthens his hand to say that Congress has his back and that our most advanced [1:19:42] chips are not for sale to our adversaries. You already responded to Ranking Member Warren earlier that you [1:19:49] support the AI Overwatch Act, but can you talk more about how ensuring that China doesn't gain access to [1:19:56] leading edge chips protects America's national security like what the AI Overwatch Act is designed to do? [1:20:02] Yes. Thanks, Senator, and thanks for leading on that, working with Chairman Mass in the House and others, [1:20:09] including your Democratic colleagues on it. The importance, as you said, is partly to codify [1:20:18] some of the Trump administration's own policy, that we're not going to send our leading edge chips [1:20:21] to China, and then putting in a structure for, I think, a really important enhancement of the [1:20:28] relationship between the Congress and the executive branch on how we do exports of these most critical [1:20:34] technologies, because like you said, they're fundamental for military purposes, for all sorts of [1:20:39] security and intelligence and human rights purposes, and they're fundamental to the strategic [1:20:44] competition broadly between the US and China, our tech companies and theirs all around the world. [1:20:49] And some of the critics of AI Overwatch suggest that somehow, this seems ridiculous to me, but somehow [1:20:57] that AI Overwatch or writing export controls into the law might weaken our hand. I mean, that's outrageous, [1:21:05] isn't it? I think that sounds right to me. You have also made the point today that the United States [1:21:12] writes strong export control rules that they often fail to enforce, and we continue to see smuggling [1:21:20] networks, moving billions of dollars in high-end chips to the Chinese military, much of it routed [1:21:26] through Southeast Asia. There are other export control bills like Senator Cotton's Chip Security Act. [1:21:32] Senator Ricketts has been a great leader on this issue through the Match Act, both bills that I fully [1:21:38] support and I've co-sponsored. There are common sense policies that address national security while also [1:21:44] ensuring American economic competitiveness is not harmed. What is the biggest gap in our export control [1:21:50] enforcement today? And how can this committee seek to fix it? Well, I think these bills are especially important. [1:21:56] I mean, bills that focus on smuggling, focus on the ability to enhance things like location verification on [1:22:05] chips. You know, just in the last 13, 14 months or so, we've had some influential voices, you know, argue that [1:22:15] essentially there's no such thing as smuggling. It might be physically impossible was the claim. [1:22:20] Now we've seen mounting evidence, evidence in news reporting, evidence in the indictment by the Southern [1:22:28] District of New York recently of some leadership of one of the biggest server companies in the world. [1:22:35] And there's a sense that there's a lot more out there, I think, in the intelligence record [1:22:40] and in the sort of things that our export control bureaucracy is able to see but hasn't always been [1:22:49] able to take full action on or prevent. And I think legislation that focuses on it as well as budget [1:22:56] support. But I think this is fundamentally a matter of real policy prioritization and then the budget [1:23:02] would follow that. Thank you. My time is expired. Senator Kim. [1:23:07] Yeah, thank you. Thank you all for joining up. I just have to say, I'm so glad that the chairman [1:23:13] called for this hearing. I think this is one of the most consequential topics that I've had to engage [1:23:18] on in a hearing in my time in the Senate so far. And I think that from what I've heard, there's wide [1:23:24] agreement amongst my colleagues about the importance for export controls and for engaging. I thank Senator [1:23:30] Ricketts for working with me on the Match Act, Senator Banks and others that I've had a chance to work on. [1:23:35] But I think what I'm trying to hit home here, Mr. Feith, I'd like to just connect with you on this, [1:23:40] because certainly the administration has taken some actions. But how important is it that we push [1:23:46] forward on this legislation and try to create an enduring and comprehensive framework? How important [1:23:52] is that for us to be able to take up here in Congress? I think very important. There's a really [1:23:57] important role for statute that lasts. Because I would say that the temporary nature, if the idea is that [1:24:03] these export controls could very well shift from administration to administration, how does that [1:24:09] affect how allies might engage or how our adversaries might engage or how business might engage? [1:24:15] I think it's hard to generalize. I mean, we've got qualities of our political system that are designed [1:24:21] for steadiness and for long-term codification. And we've got aspects where the discretion of an executive [1:24:28] is important. And that's part of why we have cycles in our politics. So I think some of that [1:24:33] is natural and appropriate. But I think to your fundamental point, Congress has an enormous role [1:24:39] in this part of this national security challenge. This committee, of course, has oversight of export [1:24:48] controls. And there are major opportunities that I think sponsors on this committee and elsewhere in [1:24:53] the Congress have pointed out to really bolster national security, strengthen our export control [1:24:58] rules and enforcement capability, and kind of assert the Congress's role in ways that would be lasting [1:25:04] on this sort of nicely interlocking set of aspects of this national security problem. [1:25:09] And I hope that we can work together to build that long-term resiliency and that direction for our [1:25:14] country. And look, given the speed in which the technology is advancing, as well as sometimes the lag in [1:25:21] which it takes to implement law and rulemaking, would you say that it's important for us to move [1:25:26] as soon as possible on this? Yes, sir. And when we're talking about aligning with our allies, [1:25:30] you know, we certainly talked about in terms of what's going to have the strongest impact to impede [1:25:35] and stop gaps from forming that our adversaries can exploit. But I would also say that it probably also [1:25:42] has a benefit of trying to level the playing field for U.S. companies in terms of how they're engaging [1:25:48] abroad as well. Yes. And the level playing field is especially important when it comes to the [1:25:52] question of so-called allied backfilling. You know, cooperation with our allies is really very [1:25:58] important, including so that we don't have companies from important allied countries that [1:26:04] aren't subject to the same types of national security restrictions that U.S. companies are subject to, [1:26:08] because these are national security concerns vis-a-vis China that we share together. [1:26:12] Chair, I just wanted to say that there's been strong support for the Match Act for these different [1:26:19] efforts we're doing from outside experts, industry stakeholders, national security experts. And so [1:26:25] to that end, I ask for unanimous consent to place a letter of support for the Match Act signed by a [1:26:30] broad-based coalition of stakeholders into the record. Without objection. Mr. Flynn, I don't [1:26:38] necessarily have a question here, but I just wanted to thank you for bringing up the importance of [1:26:42] supporting BIS and making sure that as we're pushing forward on the HIS expert controls that we [1:26:47] have the effort to be able to come behind that. And to my colleagues here in the committee, you know, [1:26:52] I'm leading the appropriations effort to try to increase funding to BIS. I hope that that's something [1:26:57] that we can continue to push on together. Dr. West, I wanted to just talk to you. One thing that I hear a [1:27:03] lot of concerns about, you know, especially from seniors and others, is about fraud and what might happen [1:27:10] next when it comes to AI's capacity to just amplify and magnify that so far beyond anything we've seen [1:27:17] before. What are the practical steps that we should be looking at about fraud from a government and [1:27:22] industry and financial institutions section? Yeah, I'm glad that you brought this up because we are [1:27:28] seeing a pattern where scams and fraud are on the rise, particularly with the off-the-shelf and [1:27:33] inexpensive availability of generative AI tools. So there's an ecosystem around fraud where law [1:27:40] enforcement and social platforms and financial institutions and tech companies all play a part. [1:27:45] I think that we need to place our greatest emphasis on the platforms and financial institutions that have [1:27:52] the greatest insights through the amount of data that they're able to collect to identify scams. And to [1:27:59] stop them sort of means that we need to be like protecting against them occurring in the in the first place. [1:28:04] So we should be requiring more of those institutions as the starting place. Yeah, thank you so much. [1:28:10] And I just want to end by just again reiterating some of my colleagues on the other side of the aisle. [1:28:14] I hope we can move forward on a markup on the Match Act and other important pieces of legislation in this committee. [1:28:20] And with that, I'll yield back. On behalf of the chairman, Senator Blunt Rochester. [1:28:24] Thank you, Ranking Member Warren. I have to say, I literally have so many notes on my dais here that I [1:28:34] don't even know where to begin because this issue is just so consequential to our country. So thank you [1:28:40] so much for your thoughtful testimony as well. It's consequential to our economy, to our health, [1:28:46] to our national security. And over the past year, I asked Chairman Powell when he was here, I asked [1:28:56] Chairman Warsh about the impact that AI is going to have on our workforce. And it was interesting to me [1:29:02] because what literally came out of it was just the uncertainty. Mr. Reinhardt, you even mentioned data [1:29:09] and the data that we collect or don't collect. We had the BLS direct nominee before us yesterday and [1:29:14] another committee. And from my time in the House to being here in the Senate, in the House, I started [1:29:20] a Bipartisan Future Work Caucus. And that was almost a decade ago. And now it is literally here [1:29:26] before us. And I can say I get to look at it from that lens. And also, I've got bipartisan legislation [1:29:32] right now with Senator Sheehy, the AI Fraud Accountability Act. I've looked at it from AI literacy [1:29:39] to help prepare our families. And Dr. West, I think you put it so perfectly that we need to put the [1:29:46] needs of the public at the center of AI policy, the needs of the public. And so I want to start off [1:29:54] by saying, as you said, there's a lot of feelings out here about this. People are either really, [1:29:59] really excited or really, really scared. They're excited about what they maybe can do from shopping or [1:30:07] decorating an office, maybe mine, or not so much. But they're also afraid that they're going to lose [1:30:15] their jobs and that this will have a detrimental impact. And so many news reports recently have come [1:30:22] out about new entrants into the workforce, whether it's college students, women in particular, and people [1:30:28] of color disproportionately being impacted by or and exposed to occupations most vulnerable to AI. [1:30:36] So I'm concerned that the AI is moving much faster than we are. And we're not taking advantage [1:30:41] of the disruptions or the opportunities. So Dr. West and Mr. Reinhardt, if you could pick one or two [1:30:48] workforce indicators that Congress should be closely watching, what would they be? I'll start with [1:30:53] you, Dr. West. Sure. So obviously, we want to be looking at indicators like layoffs as a frontline [1:31:04] measure and trying to understand how that's maybe being disproportionately felt across the economy. [1:31:09] I'd also though be looking at things like employment dislocation. We've seen, for example, a number of [1:31:17] jobs being restructured so that they're sort of flexible or self-employed. And that also means that [1:31:23] those workers don't have access to the same kinds of social safety net measures. And they're also [1:31:29] sort of dually feeling the tax burden. So that's another place that I'd be looking closely. [1:31:33] Thank you. Mr. Reinhardt. Yeah, I would I would be focusing on the unemployment rate within the AI [1:31:38] affected industries, right? There has been a whole bunch of these indices that have tried to understand [1:31:43] job loss, potential job loss and dislocation. And I will say that right now, we're really only seeing [1:31:49] it with, you know, entry level workers, people that are in the 20 to 24 range, 20, 25 range, [1:31:54] rather. And I think there's actually a very good reason for that. You know, if you're young, [1:31:58] you don't have a lot of experience, you don't have skills, you can't show to employers that you're [1:32:02] skilled, or, you know, that you have the kinds of things for it for a job. And so my sense is what's [1:32:08] happening right now is that the market is becoming very noisy. You know, we talk about this as [1:32:12] as a problem in matching between employer and employee. And so those are the sorts of things [1:32:18] that I'd be looking towards is really the unemployment in the AI affected industry and kind of the noisiness [1:32:22] of the of the market. I will follow up with the two of you and the other panelists on the data that [1:32:30] we need. And that's required. We've heard a lot of promise that AI will lower costs and improve [1:32:37] affordability. Dr. West, what factors would demonstrate that those benefits are actually [1:32:42] reaching consumers and families? That's a good question. I mean, I think the the things that we [1:32:50] would be wanting to look for are, you know, general inflationary indicators. I mean, honestly, [1:32:57] what we're what we're seeing is in many sectors of the economy, costs are really going up. And this is [1:33:02] part of a generalized trend toward inflation, the utility sector, I think, is one example of that. [1:33:08] Equally, we're seeing the rise of surveillance pricing where data is being leveraged against [1:33:12] people in ways that is increased the increasing prices across the shelf. [1:33:16] Thank you. I have one more question here, and I will send more for the record. [1:33:21] This is just a yes or no. Oh, great. [1:33:23] For everybody on the panel. Yes or no. [1:33:26] Chairman Scott talked about the use of some of the tools. We use them every day. I signed up for one, [1:33:33] loved it. And within less than a year, a pricing structure came out with multiple tiers. Are we at risk [1:33:44] of broadening the digital divide? Yes or no? I'm not an expert on that issue, Senator. I'm happy to take [1:33:53] it back and give you some some more thoughtful commentary. Okay, Mr. Feith. It's just a yes or no. [1:33:58] Do you think we're at risk? I'd have to duck it too, but I'll- I appreciate you saying you ducked it. [1:34:06] Thanks. Mr. Ryan. Yes, and I think there are things that we can be doing to address it. [1:34:10] Would love to follow up with you on that. Yes. Thank you. I yield back. [1:34:16] Just in time for Senator Alston Brooks to have an opportunity to share as well. [1:34:20] Thank you so much LPR for holding the floor for Senator Alston Brooks. Thank you so much. [1:34:25] We're so happy you made it. Thank you. I'm glad I made it too. So thank you, [1:34:29] Chair Scott and Ranking Member Warren for holding the committee. Thank you also to these witnesses [1:34:34] who are here today. You've heard already. We've heard the conversation about an artificial intelligence [1:34:41] and the enormous opportunity that it presents for our country. And if we get this right, [1:34:46] I believe that AI can help American businesses grow, make government and financial services more [1:34:52] efficient and strengthen our national security. And I believe that the United States should lead the [1:34:57] world in this technology. We shouldn't be afraid of innovation and we should not create a regulatory [1:35:04] environment that pushes the next generation of AI development overseas. But we also have to remember [1:35:10] that innovation is most successful when people have confidence in it. For our American families, [1:35:16] workers, small businesses, and communities, AI cannot just be happening around them or to them. [1:35:23] It has to expand opportunity for them. So that means making sure that people can trust the systems [1:35:29] that they are using, whether they're small businesses who are trying to compete, a worker learning new tools, [1:35:36] or a family applying for a loan, buying a home, or managing their finances. It also means being clear-eyed [1:35:42] about how this technology is deployed, especially in areas where the stakes are high for consumers, workers, [1:35:47] in the economy. And we also have to be honest that for many workers, this moment brings real questions [1:35:53] about job security and displacement. As this technology changes the way that Americans work, [1:35:59] we have to make sure that workers are not being left behind or pushed aside, but supported with the [1:36:03] skills and training that can allow them to use AI to their advantage. So the question for me is not [1:36:09] whether we should support AI innovation. We should. The question is how we make sure American innovation [1:36:16] remains fast, competitive, safe, secure, and broadly beneficial. So the first question I have for Mr. [1:36:23] Flynn, American advantage in AI depends in part on whether companies can build, deploy, and scale tools [1:36:30] securely, especially in sectors like finance, where trust and reliability are essential. So when you talk to [1:36:36] companies across the technology sector, what would you say they are most worried about? What are they saying [1:36:44] could slow AI leadership? Is it cyber vulnerabilities, would you say data rules, supply chain risks, or [1:36:50] something else? Thank you, Senator, for that question. I don't know if there's one thing, but certainly [1:36:57] the supply chain challenges are very real. As the demand for AI capabilities expands, that does squeeze [1:37:06] the supply chain in a number of areas. We don't always anticipate where those market adaptions may be, [1:37:13] so I think it's incumbent upon Congress, upon all of us to keep partnering closely on this challenge to [1:37:21] make sure that the supply chain issues can be, do not become a plateau for our ability to continue to [1:37:27] advance this technology. So now in high-stakes sectors like financial services, what do you think would [1:37:32] help adopt AI responsibly while giving consumers confidence that their sensitive data is being protected? [1:37:39] The cyber security challenge is real. Industry folks were encouraged by the president's executive [1:37:48] order recently on AI and cyber. It's a good first start in terms of creating the right bureaucratic lanes [1:37:55] and to make sure that there's good communication between industry and particularly affected critical [1:38:00] infrastructure sectors, including financial services. But at the end of the day, in this immediate term, [1:38:07] the information sharing challenge is going to be very real, which is why I've talked so much today [1:38:12] about the need to re-authorize the Cyber Information Sharing Act of 2015, CISA 2015, that lapses at the [1:38:20] end of this year. If it lapses, that would be an extraordinary unforced error in terms of the average [1:38:24] American's ability to trust that their systems are being well defended. [1:38:30] Now, lastly, to Dr. West. So I want to turn to public trust, which I see as an indispensable part [1:38:38] of innovation. So if people, for example, can't trust AI systems, they just simply won't use them. [1:38:43] If small businesses don't understand the rules, then they'll hesitate to adopt them. And if families [1:38:47] are worried that AI is deployed in such a way that they cannot see, understand, or challenge, and we've built [1:38:54] a system that can't be used. So what would it look like to build trust in a way that continues to [1:38:59] support innovation so that the American people are able to take advantage of the opportunities [1:39:04] that these new technologies provide? [1:39:06] I think trust will come when we have the right kind of sort of institutional measures in place [1:39:14] that make sure that these systems are being adequately tested and vetted before they're, you know, [1:39:20] before they're impacting people or causing harm. We don't have, you know, baseline mechanisms, [1:39:26] especially in like these critical infrastructure areas that are requiring the systems to be [1:39:33] evaluated if they're fit for purpose, if their efficacy, their safety. And I think that, you know, [1:39:39] focusing in there, making sure that this is happening before systems are entering the market is the right [1:39:45] place. Otherwise, we're kind of left for companies to be grading their own homework. [1:39:49] Great. Thank you. [1:39:51] Well, thank you all for being present with us today. It's been quite informative, [1:39:54] and we really appreciate this hearing is adjourned.