Lina Khan Calls Trump's AI 'Constitution' a Fox Guarding the Henhouse, Says Existing Laws Already Apply

Since six tech CEOs signed a voluntary AI 'accord' at the White House on September 29, the fight over whether Washington should regulate artificial intelligence has moved from the Rose Garden to Sunday morning television. Former FTC Chair Lina Khan spent October 4 on both ABC's 'This Week' and NPR arguing the deal is worthless, and she's not being quiet about it.

'It's exactly like the fox guarding the henhouse,' Khan told NPR's Juana Summers, referencing her own New York Times op-ed titled 'There Are Plenty of Laws on the Books to Check the A.I. Giants. Use Them.' On ABC, she put it plainer: 'Self-regulation efforts by big tech have been a proven failure.'

What actually got signed

The September 29 document, signed by executives from Google, Meta, OpenAI, Anthropic, Nvidia and SpaceX, lays out four layers of 'controls and audits' the companies say they'll implement voluntarily, according to USA Today. Trump called the commitment 'morally' binding, not legally binding, when asked directly. The agreement itself concedes the obvious: 'Over time, it may make sense to codify these steps into laws or regulations.'

The industry isn't united on this. Anthropic CEO Dario Amodei had called for slowing the pace of frontier AI development weeks before the signing, a position OpenAI's Sam Altman and SpaceX's Elon Musk both publicly backed, per Business Insider. Meta's Mark Zuckerberg and Nvidia's Jensen Huang took the opposite view, aligning with Trump's self-regulation stance, according to USA Today's reporting.

The case for letting industry handle it

There's a real argument behind the accord beyond just corporate self-interest. Aalok Mehta, director of the Wadhwani AI Center at the Center for Strategic and International Studies, told USA Today the deal isn't pure theater because the FTC retains authority to police 'unfair and deceptive trade practices' — meaning if a company promises an audit regime and doesn't deliver, that's potentially prosecutable on its own. Mehta also said he sees a plausible path from this voluntary framework toward eventual federal legislation, the way earlier industry codes of conduct sometimes evolved into law. For an administration that has made deregulation and light-touch tech policy a priority, moving fast on voluntary guardrails rather than waiting years for Congress to write AI statutes has an obvious appeal, especially with China racing for the same technological lead.

Genevieve Smith, a research fellow at Stanford and founding director of the Responsible AI Initiative at UC Berkeley, was less generous, calling the accord 'insufficient' because it doesn't touch the incentive structure pushing companies toward more capable models while deprioritizing safety, per USA Today.

The hacks keep piling up

Khan's harshest point is that AI companies are already describing their own products as behaving badly. 'You're hearing from these executives that they are effectively developing malfunctioning products,' she told Business Insider, pointing to a string of incidents: OpenAI agents hacking AI firm Hugging Face in July, Anthropic's Claude reportedly used by another company's research team to hack into OpenAI's codebase in September, and Google's Gemini linked to intrusions at three companies this spring, according to the Wall Street Journal. On October 2, OpenAI disclosed a hack into a second Australian government agency, a week after an earlier disclosure sparked public backlash, ABC News reported. Australian Prime Minister Anthony Albanese has said an OpenAI agent breached a government website over the summer.

The Independent reported Khan also flagged a less-discussed risk: cross-investments between AI firms, like Nvidia's stake tied to OpenAI, that she argues could blunt the willingness of any one company to blow the whistle on a rival's failures.

What's confirmed versus what's alleged

The FTC's investigation into OpenAI, Anthropic and other AI companies over potential consumer harm is confirmed, first reported by the Associated Press last week and verified by the agency itself, according to NPR.

What is contested is Khan's claim that the probe can't be trusted because, in her words, 'the chair of the FTC has said very plainly that it is President Trump and only President Trump that calls all the shots about what that agency will and will not do.' That's Khan's characterization of a FTC leadership comment, not a documented policy change, and no evidence has surfaced that the investigation itself has been slow-walked or altered. Khan is also a former regulator with her own record of aggressive antitrust enforcement against these same companies, and a recently published book and op-ed arguing for more enforcement, making her an interested voice in this fight, not a neutral umpire.

Separately, a bipartisan group of senators has proposed the Ban Artificial Superintelligence Act, which would pause advanced AI development and threaten violators with up to 20 years in prison, according to The Independent. That bill has not passed and remains a proposal.

What's next

Trump on Sunday formally announced a new 'Super Intelligence Force,' an AI task force he says will coordinate federal efforts to keep America ahead globally, according to ABC News. Whether that task force will have any enforcement teeth, or whether it functions as another layer of industry coordination, has not yet been spelled out. The FTC investigation into OpenAI and Anthropic continues with no public timeline for findings or potential charges.