Dan Hendrycks, executive director of the Center for AI Safety, at TechCrunch Disrupt in October 2025. Image: TechCrunch / Wikimedia Commons, CC BY 4.0, cropped

Ninety-one scientists, lawyers and tech figures, from Steven Pinker to Hugging Face co-founder Thomas Wolf, have put their names to a declaration that “Anthropic is not a safe AI company”. The MisAnthropic Statement, organised by the Center for AI Safety (CAIS), went up on Wednesday, hours before Anthropic is due to price the share sale that could value it at up to $1 trillion.

What the statement says

The statement is three sentences long. It opens: “Anthropic warns that recursively self-improving AI could concentrate power and escape human control, yet races to build it first.” It goes on to say the company “uses safety branding to court investors, employees, and the public”, “even as it has quietly weakened its safety pledges and lobbied for self-serving regulation.” It ends by calling on readers to “reject its reckless attempt to amass power in the name of the greater good.”

In a press release, CAIS sets out what it calls its key evidence. It says Anthropic weakened its Responsible Scaling Policy in February, “including fully abandoning its unilateral commitments to pause”, and pitches recursive self-improvement to public-market investors as “one of its core competitive advantages”. It also says Anthropic lobbied to weaken California’s SB 1047, New York’s RAISE Act and the AEGIS Act, and “privately advocated for federal pre-emption of state AI laws”. These are the coalition’s claims. Last month a leaked Anthropic draft bill that would override parts of state AI laws drew similar fire from safety campaigners, and Anthropic said it would revise it. Outside evaluator METR also said this month that Anthropic is improving its models about twice as fast as before, mostly thanks to its own AI.

Hendrycks, who runs CAIS:

You cannot credibly claim to be safe while weakening your own previous safety commitments and asking the public for tens of billions of dollars to build recursively self-improving AI.

Dan Hendrycks, executive director, Center for AI Safety

Safety campaigners, accelerationists and Fields medallists

Alongside safety figures such as Max Tegmark of the Future of Life Institute, Nate Soares of the Machine Intelligence Research Institute and PauseAI’s Holly Elmore are accelerationists who want AI built faster: Guillaume Verdon, the Extropic founder known online as Beff Jezos, and George Hotz of the tiny corp. Igor Babuschkin, a co-founder of Elon Musk’s xAI, a direct Anthropic rival, has also signed.

The list includes Nobel economist Oliver Hart, Nobel physicist Sheldon Glashow, Turing Award winner Judea Pearl, Fields medallists Peter Scholze, James Maynard and David Mumford, Harvard law professors Laurence Tribe and Lawrence Lessig, Go co-creator Rob Pike, Gary Marcus, Tristan Harris and Ed Newton-Rex. So did Phil Aroneanu, founder of Irreplaceable; Irreplaceable Education Fund is one of four nonprofits suing Anthropic and four other labs to freeze frontier AI.

Tegmark wrote on X that he signed “because I believe that Anthropic is in many ways the AI company doing the most harm.” Stephen Casper, an assistant professor at Harvard Kennedy School, accused the company of “recklessness and serial dishonesty”.

‘Unable to sign without risking relationships or future funding’

Hendrycks says the list would be longer if people felt free to sign:

He has not named the groups. Wolf, whose name sits near the top of the page, said he heard the same from “entrepreneurs, executives and researchers”. He also had doubts about the wording:

I don’t fully agree with the wording. I wouldn’t have used “reckless”, and I didn’t know the name of the statement when I signed, which isn’t my style.

Thomas Wolf, co-founder and chief science officer, Hugging Face

He signed anyway, saying Anthropic is “on track to be the most powerful company on earth” and needs openness “close to, or maybe even higher than, that of a sovereign government”.

Timed for the IPO

Anthropic is expected to set its final share price on Wednesday evening, with Nasdaq trading under ANT due on Thursday, November 19. The company set a range of $70 to $78 a share last week, and reports since say demand has run well ahead of the shares on offer. Anthropic hasn’t commented on the statement.

Why it matters

Anthropic’s safety reputation is part of what it sells to investors, staff and governments, and this challenge comes from inside the safety movement as well as from rivals, on the day its price is set. If Hendrycks is right that others stayed quiet for fear of losing funding, holding AI labs to account gets harder.

Sources: The MisAnthropic Statement and signatories (November 18, 2026); Center for AI Safety press release via Business Wire (November 18, 2026); X posts by Dan Hendrycks, Thomas Wolf, Max Tegmark and Stephen Casper (November 18, 2026).

Latest Anthropic news

More Anthropic news