
Illustration
Companies · Updated 3 min read
Anthropic's safety claims disputed by 89 signatories on eve of IPO
A day before Anthropic's IPO, 89 scientists, legal scholars and tech founders published a statement on 18 November saying Anthropic “is not a safe AI company”. Signatories include Steven Pinker, Max Tegmark, George Hotz and Hugging Face co-founder Thomas Wolf.
Key facts
- On 18 November 2026 the Center for AI Safety published the “MisAnthropic Statement”; its website listed 89 names that evening.
- Signatories accuse Anthropic of dropping a unilateral pause commitment in February and lobbying against AI safety bills.
- The Anthropic IPO is planned for 19 November 2026, with a price range of $70 to $78 a share.
Researched and written with AI, to Luigi's brief, positions and editorial rules. Automatically checked against the sources. How I work →
The text is titled the “MisAnthropic Statement”. It says Anthropic warns that self-improving AI could concentrate power and escape human control, yet races to build it first. The company uses its safety branding to attract investors, employees and the public, has quietly weakened earlier commitments and has lobbied for rules that suit it, according to the statement. Its conclusion: Anthropic is not a safe AI company.
The statement was organised by the Center for AI Safety (CAIS), a San Francisco non-profit. The press release went out on Business Wire at 11:14am ET on 18 November. When the statement's website was checked that evening it listed 89 names.
Who signed
Named signatories include Harvard psychologist Steven Pinker, economics Nobel laureate Oliver Hart, Turing Award winner Judea Pearl, Harvard law professors Laurence Tribe and Lawrence Lessig, MIT physicist Max Tegmark and Fields medallists Peter Scholze, James Maynard and David Mumford. Industry signatories include Hugging Face co-founder Thomas Wolf, xAI co-founder Igor Babuschkin, comma.ai founder George Hotz and former Y Combinator president Geoff Ralston.
The mix is unusual. AI safety researchers sit alongside open-source advocates and Guillaume Verdon, founder of the “effective accelerationism” movement, which wants AI developed as fast as possible. CAIS director Dan Hendrycks said in the release that the signatories all see the danger of a company that convinces itself its own dominance is the greater good.
The allegations
Anthropic weakened its Responsible Scaling Policy in February, dropping its unilateral commitments to pause development, CAIS says. The company also lobbied to weaken California's SB 1047, the RAISE Act and the AEGIS Act, and privately argued for federal law to pre-empt state AI rules, according to CAIS. It adds that Anthropic pitches recursive self-improvement to public-market investors as a core competitive advantage.
“You cannot credibly claim to be safe while weakening your own previous safety commitments,” Hendrycks said. The release includes no response from Anthropic, and the claims are the signatories' own account.
The Anthropic IPO is planned for 19 November, with a price range of $70 to $78 a share (timeline of the Anthropic IPO). In its amended prospectus, Anthropic says it has met its own threshold for AI that sharply speeds up the company's research, Yahoo Finance reported. The company says that calls for more safeguards, not a halt. Its first prospectus had already described how AI could soon sharply accelerate its own development.
Analysis
Safety is the core of the Anthropic brand. The company calls itself an “AI safety and research company” and puts safety at the centre of its public image. From Thursday, shareholders, analysts and critics will test that promise in public, every day. This statement is the first test, and it does not come from one camp: when safety researchers and accelerationists sign the same text, it is hard to wave away as factional infighting.
Whether the claims hold up is an open question; the signatories have their own agendas, and Anthropic has not yet replied. For companies using Claude, there is still a practical lesson: a vendor's safety commitments belong in the contract, not the pitch deck. It makes sense to record which version of the vendor's policies applied at signing and to agree on notice when they change. OpenAI is currently showing one model of outside scrutiny with external assessors who get access but do not write the report alone. Trust is a brand's most valuable asset, and on the stock market it will now be priced daily.
Update, 19 November: Anthropic priced its IPO at $83 a share, above the range, CNBC reported: Anthropic prices IPO at $83 a share, above its range
Got AI news? Submit news →
FAQ
What is the MisAnthropic Statement?
A statement published by the Center for AI Safety on 18 November 2026. It accuses Anthropic of using safety as marketing and of weakening earlier commitments.
Who signed the MisAnthropic Statement?
The list showed 89 names on 18 November, including Steven Pinker, Oliver Hart, Judea Pearl, Max Tegmark, Lawrence Lessig, Thomas Wolf, George Hotz and Guillaume Verdon.
When is the Anthropic IPO?
It is planned for 19 November 2026, with a price range of $70 to $78 a share.
Source
Center for AI Safety: Anthropic is Not a Safe AI Company
More sources
- MisAnthropic Statement – Text der Erklärung und Unterzeichnerliste (89 Namen beim Abruf 18.11. abends).
- Business Wire (via FinancialContent) – Pressemitteilung 18.11.2026, 11:14 Uhr EST, Unterzeichner mit Funktionen und Zitate.
- Yahoo Finance – 11.11.2026: geänderter Prospekt (S-1/A), Schwelle erreicht, Preisspanne 70–78 $.



