Anthropic tells employees at its San Francisco office to work from home after it received a notice saying…

Anthropic asked staff at its San Francisco offices to work from home on Monday and Tuesday this week, acting on a warning that the guards who staff those buildings might walk out. The AI company received that notice last week from Allied Universal, the contractor that supplies its security personnel, according to Business Insider, which first reported the story.There is a complication. The union representing those guards says no strike is on the table. The Service Employees International Union told Business Insider it has not called a strike authorisation vote and has issued no strike threat for this week, which leaves Anthropic’s shuttered San Francisco offices looking like a response to something that has not actually happened.

Anthropic is not at the bargaining table, but a security strike would still shut its offices

SEIU has been negotiating with Allied Universal and other California security firms since April. It is pushing for higher wages, better healthcare coverage and broader job training for the thousands of security workers it represents across the state. Anthropic is not a party to those talks, but as a client it absorbs the disruption if guards do stop showing up.Anthropic declined to comment on the work-from-home instruction. Allied Universal would not say what it told the company. A spokesperson said the firm continues to monitor the situation and stays focused on safe, reliable operations.

Security threats against AI staff explain why Anthropic emptied its offices over a maybe

The timing is awkward for an employer that has been nudging staff back in. Anthropic’s job postings describe a hybrid setup where employees are expected in office at least 25 percent of the time, so two remote days are a small but visible dent in that policy.The caution also fits a wider pattern. Security spending across the tech industry has climbed sharply in recent years, and the Wall Street Journal has reported that both Anthropic and OpenAI have dealt with threats against their employees. Anthropic is also in a sensitive stretch commercially, having filed a confidential S-1 draft in June ahead of an expected IPO, with secondary-market investors valuing it at roughly $1.5 trillion.The guards may never walk out at all. Anthropic apparently decided it would rather keep its offices empty for two days than find out the hard way.