A company nine days into the quiet period before a potential trillion-dollar IPO did something unusual on Wednesday: it asked the government for the power to stop companies like itself.

Anthropic announced an initial $200 million commitment to research AI's impact on jobs and the economy, paired with two policy proposals — one on catastrophic risk, one on labour-market disruption — while CEO Dario Amodei published a personal essay expanding on what he thinks governments owe the people AI displaces.

Regulate us like aviation

The catastrophic-risk proposal is the eye-catching one. Anthropic argues the government should be able to "block or deter" the rollout of AI models that "pose a significant risk of catastrophic harms." Amodei's analogy is aviation: AI models, he wrote, should go through technical testing and auditing with the rigour the Federal Aviation Administration applies to aircraft before they're allowed to carry passengers.

That is a frontier lab proposing a regulator with the power to ground its own products. Cynics will note it's also a frontier lab — one whose models sit at the top of capability leaderboards — proposing a compliance bar that would weigh heaviest on smaller rivals. Both readings can be true at once; that tension is the policy fight of the next two years in miniature.

The labour half: bigger, longer, different

The second proposal deals with what AI does to work. Amodei's framing is notably blunt for a CEO whose revenue depends on enterprises adopting exactly this technology: AI could produce much larger disruptions to the labour market than previous technological shifts — and those disruptions could last longer.

The line that will travel: "The key challenge in such a world won't be incentivising growth, but finding a way for everyone to share in the benefits." In accompanying interviews, Amodei went as far as suggesting that something like a universal basic income might become necessary as AI displaces jobs.

The $200 million research commitment is the concrete piece — funding work on measuring AI's real economic impact, which is currently more anecdote than data. (Economists have spent 2026 arguing about whether AI is already adding hundreds of billions to the economy without showing up in official statistics; nobody actually knows.)

Why now is the interesting question

The timing does a lot of work here. Anthropic confidentially filed for an IPO on 1 June at a $965 billion post-money valuation. Nine days later it's publishing proposals about catastrophic risk, job destruction, and wealth-sharing.

One read: this is a values-led lab using its loudest moment to set the policy agenda before going public makes that harder. Another: it's pre-emptive reputation work — getting on the right side of the labour question before the IPO roadshow forces awkward conversations about what a $1 trillion AI company means for everyone who doesn't own shares in one.

Either way, the substance matters. A major lab has now put on paper that governments should hold the power to block frontier model deployments, and that the labour disruption is real enough to start funding the measurement of it. Those are markers that don't get walked back easily — including by Anthropic itself, whichever side of the regulatory table it ends up on.