Scientists and lawyers publish a statement calling Anthropic "not a safe AI company"
The one-paragraph statement went up the day Anthropic priced its IPO; signatories span safety researchers, open-source advocates and accelerationists, and the Center for AI Safety listed its evidence.
- Culture & impact
- Safety & alignment
- Minor
On 18 November, the day Anthropic priced its IPO, the Center for AI Safety (CAIS) said it had joined a coalition signing the MisAnthropic Statement, published at misanthropicstatement.com. The statement, in full:
Anthropic warns that recursively self-improving AI could concentrate power and escape human control, yet races to build it first. 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. Anthropic is not a safe AI company, and we must reject its reckless attempt to amass power in the name of the greater good.
CAIS’s release lists “prominent signatories” with their affiliations, among them the psychologist Steven Pinker, the computer scientist Judea Pearl, the physicist Max Tegmark, the law professor Lawrence Lessig, Hugging Face co-founder Thomas Wolf, xAI co-founder Igor Babuschkin and CAIS executive director Dan Hendrycks. CAIS described the group as safety researchers, open-source advocates, AI ethicists and accelerationists. Hendrycks said: “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.” The statement’s website invites further signatures, so any count depends on the date it was taken.
CAIS’s “key evidence” has three strands. Anthropic warns of recursively self-improving AI while pitching it to investors as an advantage, which CAIS ties to the company’s S-1 amendment. Anthropic markets itself as a safety company to recruit and raise money. And in February it weakened its Responsible Scaling Policy, “including fully abandoning its unilateral commitments to pause”, and lobbied against California’s SB 1047, New York’s RAISE Act and the AEGIS Act.
Those are CAIS’s characterisations. Anthropic’s own account of the February policy rewrite said it separated unilateral commitments from industry-wide recommendations and added risk reporting. The release carries no Anthropic reply, and none was found in the sources opened. The statement came as Anthropic’s Nasdaq listing approached; this entry records no measured effect on the offering.
In the commentary
What people were saying around this time — external links, from the record's commentary rail.