The Implications of Trump's Win for AI

0:00Today on the AI Daily Brief, the AI implications of last night's elections. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show notes.

0:18Hello, friends. Yesterday was, of course, a very consequential election where, frankly, we have a little bit more certainty around what happened than I thought we were going to. And so today what we're going to do is sort of a long-read style show, although we're going to do it in the immediate aftermath as people try to make sense of what the implications are specifically for artificial intelligence. So what we have are one full op-ed plus two segments of pieces all about AI in the elections and what happens next. And the first is a segment from Kevin Roos in the New York Times from a larger piece called What a Trump Victory Means for Tech.

0:50The section is called AI Progress Accelerates. By the way, of course, these will be read by the 11 Labs version of myself. And so here's the first. AI Progress Accelerates. Neither Mr. Trump nor Ms. Harris said much about artificial intelligence on the campaign trail. But it's a safe bet that AI progress will continue under a second Trump administration and might even speed up. Some of the tech elites who supported Mr. Trump, including the venture capitalist Mark Andreessen, are associated with the accelerationist wing of the AI movement and have opposed any AI regulations that could slow down the industry.

1:26To be honest, I'd bet against Mr. Trump spending much time thinking about AI at all. He might delegate it to Mr. Vance, who does seem to have some interest in the subject But to the extent Mr. Trump's views on AI will be shaped by the people in his orbit It's likely to be in the direction of encouraging American tech companies to race ahead of rivals in China and elsewhere And removing regulatory roadblocks that might get in their way Interestingly, Mr. Musk is a bit of a wild card here He runs an AI company, XAI, that would benefit from light-touch regulation But he also worries about existential risk from AI and supported a controversial California bill that would have imposed safety standards on AI models, which many AI companies opposed.

2:07All right, back to non-AI NLW here. This one didn't really have all that much in it. Effectively, Roos' argument is that Trump is unlikely to spend all that much time thinking about AI. That J.D. Vance might, given that he's a former VC. The most interesting thing about this piece is the question of Elon Musk, where Roos has rightly identified that on the one hand he has the existential risk worries, but on the other hand, he could benefit from light-touch regulation, I think Roos is right to call out that that could create some interesting tension to the extent that Musk has a seat at the policy table when it comes to AI, although there's no guarantee that that's actually how it's going to play out.

2:42All right, next up, we're going to read a broader piece, or AI me is going to read a broader piece, by Kyle Wagers in TechCrunch called What Trump's Victory Could Mean for AI Regulation. A grueling election cycle has come to a close. Donald Trump will be the 47th president of the U.S. and with Republicans in control of the Senate and possibly the House, his allies are poised to bring sea change to the highest levels of government. The effects will be acutely felt in the AI industry, which has largely rallied against federal policymaking. Trump has repeatedly said he plans to dismantle Biden's AI policy framework on day one and has aligned himself with kingmakers who've sharply criticized all but the lightest touch regulations.

3:21Biden's approach. Biden's AI policy came into force through executive order passed in October 2023. Congressional inaction on regulation precipitated the executive order, whose precepts are voluntary, not compulsory. The AIEO addresses everything from advancing AI in healthcare to developing guidance designed to mitigate risks of IP theft. But two of its more consequential provisions, which have raised the ire of some Republicans, pertain to AI's security risks and real-world safety impacts. One provision directs companies developing powerful AI models to report to the government how they're training and securing these models and to provide the results of test design to probe for model vulnerabilities.

4:00The other provision directs the Commerce Department's National Institute of Standards and Technology, NIST, to author guidance that helps companies identify and correct for flaws in models, including biases. The AIEO accomplished much. In the last year, the Commerce Department established the USAI Safety Institute, AISI, a body to study risks in AI systems inclusive of systems with defense applications. It also released new software to help improve the trustworthiness of AI, and tested major new AI models through agreements with OpenAI and Anthropic. Critics allied with Trump argue that the EO's reporting requirements are onerous and effectively force companies to disclose their trade secrets.

4:40During a House hearing in March, Representative Nancy Mace said they could scare away would-be innovators and impede more chat-GPT-type breakthroughs. Because the requirements lean on an interpretation of the Defense Production Act, they've also been labeled by some Republicans in Congress as an example of executive overreach. At a Senate hearing in July, Trump's running mate, J.D. Vance, expressed concerns that preemptive overregulation attempts would entrench the tech incumbents that we already have. Vance has also been supportive of antitrust, including efforts by FTC Chair Lina Khan, who's spearheading investigations of big tech companies' acquihires of AI startups.

5:15Several Republicans have equated NIST's work on AI with censorship of conservative speech. They accuse the Biden administration of attempting to steer AI development with liberal notions about disinformation and bias. Senator Ted Cruz recently slammed NES's woke AI safety standards as a plan to control speech based on amorphous social harms. When I'm re-elected, Trump said at a rally in Cedar Rapids, Iowa last December, I will cancel Biden's artificial intelligence executive order and ban the use of AI to censor the speech of American citizens on day one. Replacing the AI EO So what could replace Biden's AI EO?

5:50Little can be gleaned from the AI executive orders Trump signed during his last presidential term, which founded national AI research institutes and directed federal agencies to prioritize AI R &D. His EOs mandated that agencies protect civil liberties, privacy, and American values in applying AI. help workers gain AI-relevant skills, and promote the use of trustworthy technologies. During his campaign, Trump promised policies that would support AI development rooted in free speech and human flourishing, but declined to go into detail. Some Republicans have said that they want NEST to focus on AI's physical safety risks, including its ability to help adversaries build bioweapons, which Biden's EO also addresses.

6:32But they've also shied away from endorsing new restrictions on AI, which could jeopardize portions of NIST's guidance. Indeed, the fate of the AISI, which is housed within NIST, is murky. While it has a budget director and partnerships with AI research institutes worldwide, the AISI could be wound down with a simple repeal of Biden's EO. In an open letter in October, a coalition of companies, non-profits, and universities called on Congress to enact legislation codifying the ASE before the end of the year. Trump has acknowledged that AI is very dangerous and that it'll require massive amounts of power to develop and run, suggesting a willingness to engage with the growing risks from AI.

7:11This being the case, Sarah Kreps, a political scientist who focuses on U.S. defense policy, doesn't expect major AI regulation to emerge from the White House in the next four years. I don't know that Trump's views on AI regulation will rise to the level of antipathy that causes him to repeal the Biden AIEO, she told TechCrunch. Trade and state rulemaking Dean Ball, a research fellow at George Mason University, agrees that Trump's victory likely augurs a light-touch regulatory regime, one that'll rely on the application of existing law rather than the creation of new laws. However, Ball predicts that this may embolden state governments, particularly in Democratic strongholds like California, to try to fill the void.

7:51State-led efforts are well underway. In March, Tennessee passed a law protecting voice artists from AI cloning. This summer, Colorado adopted a tiered, risk-based approach to AI deployments. And in September, California Governor Gavin Newsom signed dozens of AI-related safety bills, a few of which require companies to publish details about their AI training. State policymakers have introduced close to 700 pieces of AI legislation this year alone. How the federal government will respond to these challenges is unclear, Ball said. Hamid Ekbia, a professor at Syracuse University studying public affairs, believes that Trump's protectionist policies could have AI regulatory implications.

8:30He expects the Trump administration to impose tighter export controls on China, for instance, including controls on the technologies necessary for developing AI. The Biden administration already has in place a number of bans on the export of AI chips and models. However, some Chinese firms are reportedly using loopholes to access the tools through cloud services. The global regulation of AI will suffer as a consequence of new controls, despite the circumstances that call for more global cooperation, ECBIA said. The political and geopolitical ramifications of this can be huge, enabling more authoritarian and oppressive uses of AI across the globe.

9:05Should Trump enact tariffs on the tech necessary to build AI, it could also squeeze the capital needed to fund AI R &D, says Matt Middlestead, another research fellow at George Mason University. During his campaign, Trump proposed a 10 % tariff on all U.S. imports and 60 % on Chinese-made products. Perhaps the biggest impact will come from trade policies, Mittelstadt said. Expect any potential tariffs to have a massive economic impact on the AI sector. Of course, it's early. And while Trump for the most part avoided addressing AI on the campaign trail, much of his platform, like his plan to restrict H-1B visas and embrace oil and gas, could have downstream effects on the AI industry.

9:45Sandra Wachter, a professor in data ethics at the Oxford Internet Institute, urged regulators, regardless of their political affiliations, not to lose sight of the dangers of AI for its opportunities. These risks exist regardless of where you sit on the political spectrum, she said. These harms do not believe in geography and do not care about party lines. I can only hope that AI governance will not be reduced to a partisan issue. It is an issue that affects all of us, everywhere. We all have to work together to find good global solutions. All right, back to real NLW here. And once again, we have a big shrug arms thrown up.

10:19I have no idea. Specifically, what Kyle is pointing out here is that what Trump has promised to not do or to repeal in the case of the Biden AI executive order is clearer than what he's promised to do when it comes to accelerating AI. I think Kyle is right to identify that one of the areas where AI is going to overlap with the presidency in a significant way is as it relates to trade policy and I think foreign policy more broadly. China policy is obviously going to be a big part of the next Trump administration, and AI is likely to be a big part of China policy. Today's episode is brought to you by Vanta.

10:55Whether you're starting or scaling your company's security program, demonstrating top-notch security practices and establishing trust is more important than ever. Vanta automates compliance for ISO 27001, SOC 2, GDPR, and leading AI frameworks like ISO 42001 and NIST AI risk management framework, saving you time and money while helping you build customer trust. Plus, you can streamline security reviews by automating questionnaires and demonstrating your security posture with a customer-facing trust center all powered by Vanta AI. Over 8 ,000 global companies like LangChain, Lila AI, and Factory AI use Vanta to demonstrate AI trust and prove security in real time.

11:30Learn more at vanta.com slash nlw. That's vanta.com slash nlw. Today's episode is brought to you by Superintelligent. Every single business workflow and function is being remade and reimagined with artificial intelligence. There is a huge challenge, however, of going from the potential of AI to actually capturing that value. And that gap is what Superintelligent is dedicated to filling. Superintelligent accelerates AI adoption and engagement to help teams actually use AI to increase productivity and drive business value. An interactive AI use case registry gives your company full visibility into how people are using artificial intelligence right now.

12:10Pair that with capabilities building content in the form of tutorials, learning paths, and a use case library. And Superintelligent helps people inside your company show how they're getting value out of AI while providing resources for people to put that inspiration into action. The next three teams that sign up with 100 or more seats are going to get free embedded consulting. That's a process by which our super intelligent team sits with your organization, figures out the specific use cases that matter most to you, and helps actually ensure support for adoption of those use cases to drive real value.

12:41Go to bsuper.ai to learn more about this AI enablement network. And now back to the show. Now, one analysis that I did think was pretty interesting is this idea that perhaps in the absence of a strong, clear federal regulatory regime when it comes to AI, state governments might intervene. First, I could absolutely see that happening. And second, if it does, I could also see that being a prompt for a second Trump White House to try to get more involved in pushing through their vision of whatever a light-touch regulatory regime will be. Finally, let's turn it over to our third segment, which was in the information in Stephanie Palazzolo's newsletter.

13:19It's called The Battle over open-source AI will be settled in Trump's presidency. Handing it back over to the Eleven Labs version of myself for this last read. Donald Trump's victory in Tuesday's presidential election has a lot of implications for AI. Some are clear. Trump has said he will repeal President Joe Biden's AI executive order, which requires companies training large AI models to share information about them with the federal government. In another key area, though, Trump's stance isn't certain. That's the hotly contested question of whether freely available, open-source AI such as Meta Platform's LAMA models are a threat to national security.

13:54The issue was put in the spotlight last Friday, when Reuters reported that Meta's LAMA models were being used by Chinese researchers to build AI tools for potential military applications. What made the report particularly problematic for Meta is that it had prohibited US government agencies and contractors the ability to use LAMA for military purposes. Although on Tuesday, Meta published a blog post saying it had dropped that prohibition. Allowing the US military to use Lama might not be enough to mollify Meta's critics on this issue. Trump is anti-China and has historically restricted exports of technology such as encryption software, drones, and semiconductors.

14:31The Biden administration increased those restrictions on technology. It's hard to imagine Trump would agree to let China gain access to state-of-the-art AI models. Already, OpenAI, Anthropic, and other proprietary software makers have said governments should regulate the most capable AI models so bad actors can't easily use them, and Anthropic has explicitly opposed open-source AI over safety concerns. Meta CEO Mark Zuckerberg, for his part, argued in a July blog post that, Our adversaries are great at espionage. Stealing models that fit on a thumb drive is relatively easy, and most tech companies are far from operating in a way that would make this more difficult.

15:07That's a tough argument to make with the national security establishment, even if he's right. Much could depend on who advises Trump on AI policy. For instance, his vice president-elect, J.D. Vance, has spoken out in support of open-source AI in the past, arguing that the technology will prevent crazy people from using AI to propagate left-wing bias. Trump has also said that he'd ask open-source proponent Elon Musk for advice on tech policy. At the end of the day, though, it's unclear how much influence the two will actually have on Trump's policy. And even if Trump does crack down on open-source AI, all hope might not be lost for Meta.

15:46After all, it could always choose to simply make its Lama models closed source. In doing so, though, Meta will be entering the tight race between proprietary model developers without two of its major advantages, the benefit Lama users receive from having full control over the model and the halo effect Meta gets from framing itself as a crusader for everyday developers. All right, back to Real Me. Once again, Stephanie here is identifying that there are geopolitical implications and that the foreign and trade policy that Trump wants to pursue vis-a-vis China is going to have a deterministic impact on how he and his White House think about artificial intelligence.

16:21What's different about where she's going with this is the open source AI implications. One of the big points of opposition to California's SB 1047 was, of course, the unintended potential consequences for open source AI, but is that going to run up against China policy? There are a lot of ways that this could play out, and some fairly significant tensions and push and pull from both sides. Now, if you are left feeling at the end of all three of these reads that we just don't know what AI is going to look like under Trump, at least not when it comes to the specifics, it's because of course we don't.

16:54What is clear though, is that this will be a major issue for this next administration. It'll be a national security issue, a trade policy issue, an economic competitiveness issue, an energy issue. And so you better believe that from here on out, I'm going to be spending a lot more time trying to understand how the broader Trump camp thinks about these things and what it might mean for this industry as a whole. For now, though, that is going to do it for today's AI Daily Brief. Appreciate you listening as always. And until next time, peace.

17:34한국수ydd