| The AI freakout in Washington has arrived. In recent days, the industry’s leaders have called for government intervention and safeguards to slow the pace of development and mitigate an array of catastrophic risks, including the creation of biological weapons, autonomous hacking, and the end of humanity altogether. |
| That all sounds pretty bad, but there is little reason to believe that anything will happen anytime soon. Over the weekend, Donald Trump and Republican House Speaker Mike Johnson threw cold water on the idea of Congress regulating the industry. It was further confirmation of what Democratic congressman Sam Liccardo of California told me back in July — that Republicans control both the White House and Congress, but “in this case, the party is reduced to Donald Trump’s whims.” |
| Last week was the 25th anniversary of 9/11 — a tragedy that arguably could have been avoided if officials in the George W. Bush White House had been responsive to warnings that they received throughout 2001 about the prospect of a terrorist attack on U.S. soil by Osama bin Laden. There were similar warning signs in the run-up to the 2008 financial crisis, which, according to the most definitive government account, was in part the fault of public officials who “ignored warnings and failed to question, understand and manage evolving risks” within the financial system. |
| AI poses even worse risks, but in this case, the government’s responsibility for a disaster would be even clearer, more direct, and less forgivable. There are two principal reasons for this. |
| The first is that we have industry leaders and advocates themselves publicly calling for aggressive government intervention, even though they stand to make extraordinary sums of money from pushing forward. There is no precedent for this in contemporary American politics, and there is no excuse for ignoring them. |
| On Sunday, Johnson — who is likely enjoying his last few months as Speaker — told CNN that Congress should be wary of getting involved in regulating AI because the “guys who are raising the alarms right now all have very different ideas on what the guardrails should be.” |
| The argument is facile at best and disingenuous at worst, since it is almost always true when the government regulates an industry that stakeholders have different views on what the proper framework should be. A properly functioning legislative body is supposed to channel and manage those competing interests, however imperfectly. |
| The second reason that an AI disaster would represent an even worse failure of government than the failures that preceded 9/11 or the financial crisis is that this is not a situation where policymakers have ignored or misunderstood the risks. To the contrary, the Trump White House has been deliberately indifferent to these risks and taken steps that have exacerbated them. |
| Last December, Trump issued an executive order on AI that declared that the AI industry “must be free to innovate without cumbersome regulation.” The order called on Congress to create “a minimally burdensome national standard” and threatened to withhold federal funding for states that passed laws regulating the industry (with several narrow exceptions). The order also directed the Justice Department to create an AI Litigation Task Force “whose sole responsibility” is to challenge state AI laws. |
| It also claimed that federal actions and regulations legally preempt the states’ efforts, but the president does not determine this. In fact, the legal framework for federal preemption disfavors preemption and limits the doctrine to situations in which Congress has either expressly or implicitly preempted state regulation in an area. Neither has plausibly happened here. |
| In March, the White House released a National Policy Framework for AI that reiterated these positions. The White House once again called on Congress “to ensure a minimally burdensome national standard” and to preempt the states. |
| The White House’s deeply unserious approach to this issue was underscored over the weekend by David Sacks, an entrepreneur and tech industry hype man with no prior government experience who served as the Trump White House’s AI and crypto czar until earlier this year. Sacks penned a long and confused social media post in which he accused industry leaders of undertaking “another bid for regulatory capture — or an election-season psyop.” |
| Among other things, Sacks told industry leaders to “stop pretending antitrust law has to be suspended so you can form a cartel.” “The easiest way not to build superintelligence,” Sacks argued, “is for you to agree not to build it.” |
| This is entirely backwards. An agreement by AI companies — competitors — to restrict their output could be illegal under antitrust law. |
| Even if an illegal cartel seems desirable in this particular instance, it is a bad idea to encourage industries to coordinate in this way. It is also foolish to assume that industry self-regulation would contain the risks that the companies are themselves imploring Congress and the White House to address. |
| The companies are free to make these decisions independently, but for obvious reasons — money — none of them want to unilaterally slow down while others push ahead. The way to bind them all, and to eliminate the collective action dilemma, is to pass industrywide laws and regulations — something that Congress used to know how to do. |
| At this point, the arguments for continued legislative and regulatory inertia are not only unpersuasive but foolish. |
| Trump’s principal objection is that China will take the lead on AI, but this is the one area where the supposed dealmaker-in-chief has no interest in trying to make a deal to prevent a potential global catastrophe. Meanwhile, conservative commentators with no experience or expertise in AI, law, or foreign policy are confidently declaring, without even the pretense of an effort, that a deal with China is unattainable. |
| The main problem with the Trump White House negotiating a deal with China on AI, however, is not the concept but the execution. |
| There is little reason to believe this administration could competently broker a deal — even if the Chinese were willing to entertain one — given its dismal record of global diplomatic failures and of antagonizing our allies all over the world in trade negotiations that have failed to achieve much of anything except higher prices for consumers. |
| There is no question at this point that the political status quo is dangerous. Democrats may take back one or both houses of Congress in November, but they will not be able to pass any legislation without buy-in from some congressional Republicans and, more importantly, the White House. |
| The main obstacle to regulating the industry right now is not Congress or — remarkably — the industry itself. It’s Trump. |
| If something truly awful happens, Trump won’t be able to say that he wasn’t warned. |
