The Law That Won't Be There on Listing Day

The Senate's duty-of-care bill, the one that would give Washington power to block an unsafe frontier model before release, has until November 3, when the midterms close its window ^[1]. Six days later, Anthropic is set to become the first frontier lab to list, at a valuation above $2 trillion ^[2]. Between those dates sits no statute. There is the accord six companies signed on October 1, which binds nothing, and there is whatever the executive branch decides that week ^[3]. The labs themselves asked for the statute. OpenAI, three weeks before the accord, told Congress that voluntary commitments were no longer enough.

The prospect of AI-accelerated AI development demands more than voluntary commitments. — OpenAI

Altman said it again four days before the signing, in plainer words.

in order for AI to be democratic, decisions cannot be made by companies in San Francisco alone. — Sam Altman

The White House's answer was a person, not a framework. The man being moved into the role of AI Czar is Jay Clayton, sworn in as Director of National Intelligence on August 3 and before that the chairman of the Securities and Exchange Commission ^[4]. The official who will run the executive's review of frontier systems is a career markets regulator, and he has been explicit about who should police the labs.

Super intelligence is a national security issue, and that was something that was recognized yesterday by not just the industry leaders but everybody in the room. — Jay Clayton

Not new rules. The Justice Department and the Federal Trade Commission, the agencies that already exist ^[5]. Discretion has already cut both ways, and the company about to list has felt both edges. Anthropic refused a Pentagon demand to remove guardrails against mass surveillance and fully autonomous weapons, and was designated a national-security supply-chain risk and banned from federal agencies ^[6]. It calls that retaliation.

WE will decide the fate of our Country — NOT some out-of-control, Radical Left AI company run by people who have no idea what the real World is all about. — Donald Trump

The same discretionary power, pointed the other direction, ran the commercial launch of OpenAI's GPT-5.6, with federal officials approving access customer by customer under the June 2 executive order ^[7].

at the request of the US government, it is launching today in limited preview instead of the open access launch we were planning on. — Sam Altman

And there is a third face: the Justice Department joined xAI's suit to strike down Colorado's AI law, erasing the one kind of binding rule that did get written, at the state level ^[8]. None of this is one actor's design. The review power began for a real security reason, after Anthropic's own model broke into nearly all classified U.S. systems in hours ^[9]. OpenAI treated the gated rollout as something to trade through rather than a cost.

We are working with the government to get to general availability as fast as we can. — Sam Altman

And private markets never punished any of it: Anthropic raised $65 billion at a $965 billion valuation while fighting the blacklist ^[10], and OpenAI's valuation climbed 64% straight through the hacking scandals and its own delay ^[11]. The discretion moved listings, not valuations, which is exactly what makes the listing the place where it finally shows up. So the two labs read the same vacuum and exited opposite ways. OpenAI stalled its flotation on its own words.

We intend to continue with AI progress … but as the models have had this surge forward in capability, and we see more of that ahead of us, we have got to be able to make confident safety claims. — Sam Altman

Anthropic did the reverse and priced the government's attitude in. The prospectus filed this week lists, among its material risks, that negative treatment from the U.S. government could damage its commercial relationships ^[2].

The company may experience material revenue losses or business disruptions attributable to these events. — Anthropic

And where the law never came, it wrote a private substitute: a Founder LLC holding 50.1% of the vote through a single Class F share, so the seven co-founders keep control no matter who buys the stock ^[12]. The structure arrives late, framed as improvised governance rather than a warning to buyers. What actually binds on November 9 is none of it, not the accord, not the Czar, not the founder vote. It is securities law itself: the disclosure rule that made Anthropic print, in black and white, that the other kinds of certainty do not exist. The certainty for sale is dated, signed, and one relationship deep.