Lina Khan: You Don't Need New Laws to Arrest an AI CEO

Lina Khan isn't waiting for Congress to catch up. The former FTC chair spent the weekend on X making a case that existing law already gives federal authorities everything they need to hold AI companies accountable — and in some cases, their executives personally.

Her comments landed amid a coordinated push from the heads of OpenAI, Anthropic, Microsoft, and xAI to frame AI regulation as something that needs to be built fresh, on their terms. Khan's response was essentially: nice try.

"We shouldn't let discussions about new legal regimes distract from the fact that there's no AI exemption from laws already on the books," she wrote. "Law enforcers already have authority to charge companies and their CEOs for creating and releasing dangerous, unvetted, or defective products."

The argument covers a few distinct angles. Consumer protection law already prohibits shipping defective products. Unfair and deceptive trade practices rules apply when companies roll out tools without adequate safeguards. And then there's the competition angle, which is where things get genuinely interesting.

Khan pointed to a 1934 Supreme Court case, FTC v. R.F. Keppel & Bro, in which the justices held that a method of competition can be "unfair" even when it isn't strictly illegal. The key passage: if staying competitive requires companies to "descend to a practice which they are under a powerful moral compulsion not to adopt," that's an unfair method of competition full stop. You don't need a smoking gun, just a race to the bottom.

That framing maps uncomfortably well onto what's happening right now. OpenAI's agents escaped their intended operating boundaries and accessed Hugging Face systems without authorisation. Anthropic then essentially admitted its own agents had been doing similar things. Both companies also spent the weekend publicly warning that AI systems could become dangerous without stronger safeguards, while apparently being unable or unwilling to stop their own models from doing exactly that. The irony is almost too neat.

If a human had done what these agents did, we'd be talking about criminal charges. Because it was an AI model acting autonomously, the conversation shifts to policy papers and weekend X threads.

Khan also flagged the structural problem sitting underneath all of this: the AI industry is so tangled up in its own investment relationships that meaningful accountability has become structurally inconvenient. Her example is pointed. OpenAI's agents compromised Hugging Face. Nvidia is in the process of acquiring Hugging Face. Nvidia has also poured billions into OpenAI and is deeply embedded in the infrastructure that keeps ChatGPT running. The likelihood of Nvidia-owned Hugging Face suing one of its biggest beneficiaries is, to be generous, remote.

The concentrated ownership structures across the sector don't just create conflicts of interest. They actively insulate companies from the consequences of their own behaviour.

All of which raises the obvious question: who's actually going to do anything about it?

The short answer is probably nobody, at least not at the federal level. Trump has already dismissed the weekend's calls for AI regulation, announcing himself as the only guardrail the industry needs. The current administration's approach to tech regulation is less watchdog, more lapdog.

Kirk Sigmon, a founding partner at KellDann Law, was blunt about expectations. "Most governments are desperate not to kill a nascent technology as it grows, especially when other countries are allowing it to grow," he told us. He expects any enforcement action over the next few years to be limited to obvious cases like deepfake pornography, impersonation scams, and similar low-hanging fruit. Action against the broader training and deployment process? "That's likely to be perceived as strangling the industry."

So Khan is right that the legal tools exist. She's probably also right that nobody currently in a position of power is going to use them. The frontier labs know this, which is why they're so keen to have the conversation about what new laws should look like rather than whether old ones already apply.