Anthropic reported a threatening Claude chat to the San Francisco Police Department after a user allegedly wrote that he had purchased an AR-15 semiautomatic rifle and had cofounder and CEO Dario Amodei in his sights, according to a police report. The user also allegedly threatened to kill everyone at the company and said he had bought a firearm to carry it out, the report states. Anthropic contacted SFPD about the alleged threats.
The incident occurred last month, and Anthropic later called SFPD about it, according to reporting by The San Francisco Standard. The Claude user, who has not been identified by the Standard because he has not been arrested or charged, later told the outlet he was just fucking around when he wrote the messages. Anthropic referred the matter to law enforcement.
A Lobby Visit and an Anthropic Employee's Reluctance
A Lobby Visit and an Anthropic Employee's Reluctance The incident wasn't the first physical scare at Anthropic's San Francisco headquarters this year. Security guards previously stopped a man in the company's lobby who said an executive would be killed, as reported by the Wall Street Journal. The Claude chat report also prompted contact with SFPD.
Santa Clara University law professor Eric Goldman said internet companies face significant liability if they fail to report a threat and a crime subsequently occurs — a legal calculus that helps explain why Anthropic escalated the Claude chat to police rather than handling it purely internally. The reported exchange involved a threat of violence.
A Wider Pattern of Threats Against AI Executives
The Anthropic case landed amid broader attention to threats against the AI industry.
That surge followed real-world violence earlier in the year. In April, OpenAI CEO Sam Altman's Russian Hill home was targeted with a Molotov cocktail, an attack Hoodline previously reported led to attempted-murder charges against a Texas man who also tried to break into OpenAI's Mission Bay headquarters. The incidents described above involve Anthropic and OpenAI executives.
Executive Protection Budgets Climb Across the Industry
The response from major tech firms has been to spend more on protecting the people at the top. Palantir, Oracle, and Salesforce all increased spending on executive protection over the past year, part of a broader corporate trend as technology firms pour capital into residential security and travel protection for their leadership. The incident has also drawn attention to security concerns involving AI-company personnel.
A Security Force Under Its Own Strain
The threat against Amodei surfaced just as the guards responsible for protecting Anthropic's own building were locked in an unresolved labor dispute. Security workers and contractor Allied Universal have not agreed on terms for a new contract since the previous one expired in June, and Anthropic told its San Francisco-based employees to work from home due to a potential security-worker strike. That instruction came even though no walkout had actually been authorized at the time — the dispute remains unresolved.
The tension over guard pay remains part of the dispute. Anthropic's own footprint has ballooned even as its guards fight for basic raises — the company now occupies more than 850,000 square feet across four buildings along Howard Street, a corridor dubbed AI Alley by local real estate brokers. According to San Francisco Planning, a proposed office and retail project was located at 881-899 Howard Street. According to San Francisco Planning, the 5M project proposed three new buildings, including a 630,000-square foot office building on Howard Street.
The provided account concerns the alleged Claude threat. As of this report, the Claude user involved in the August incident has not been identified, arrested, or charged.
