An unfair-labour-practice strike has been authorised by AI lab security workers, with union officers marching through San Francisco on Wednesday near the offices of OpenAI and Anthropic after a vote backed by members employed across some of the most prominent addresses in Silicon Valley. Whether a walkout will actually proceed, and when, remained unclear as of Wednesday, but a concrete deadline is now in place: according to Business Insider, the security officers submitted a 72-hour strike notice on Tuesday morning, planning to skip any shift beginning on or after 4:30 a.m. on Friday.
What the AI lab security workers strike dispute is really about
The workers are employed not directly by the tech firms, but by staffing agencies that provide building security for OpenAI, Anthropic, Google, Meta, and Salesforce. The union involved, the SEIU-United Service Workers West (SEIU-USWW), which represents roughly 14,000 security workers, says it has been bargaining with Allied Universal and other agencies, including Securitas and GardaWorld, over wages and benefits covering staff in the San Francisco Bay Area, Silicon Valley, and Los Angeles County.
The gulf between the two sides is wide. After 19 bargaining sessions, Allied Universal proposed a wage increase of just 25 cents an hour in 2027, with no raises for the three years that follow, according to Metaintro. That offer, put against a cost-of-living backdrop in one of the most expensive urban corridors in the United States, is central to the union’s grievance.
Christopher Yates, an Allied Universal security officer who lives in Hayward, put the union’s position plainly in an emailed statement: ‘We’re asking Allied and the other security companies to stop playing games, come to the table, and settle a strong contract with real raises, better training, and respect for the officers who keep these buildings safe.’
Representatives for OpenAI, Anthropic, Salesforce, Google, and Meta did not immediately respond to a request for comment.
Rising physical threats raise the stakes for AI lab security
The dispute unfolds against a backdrop of documented physical threats to AI company premises and personnel, which makes the condition and morale of security staff a more pointed concern than it might otherwise be. The Wall Street Journal has reported that Anthropic and OpenAI have dealt with several threats against their employees. In April, police said a Molotov cocktail was thrown into the home of OpenAI’s Sam Altman, who later said AI ‘doomerism talk’ may have contributed to the attack.
A separate incident at Anthropic’s San Francisco headquarters illustrates how physical access controls can fail even in buildings with professional security in place. On 15 April, according to the New York Post, a man entered the building by closely following behind an employee who swiped in with their badge, a tactic sometimes called tailgating. The Wall Street Journal first reported that incident.
Anthropic’s awareness of potential disruption to its security arrangements was visible even before the strike vote. Business Insider had previously reported that Anthropic asked employees to work from home last week, citing a belief that a strike by Allied Universal, the staffing firm covering its San Francisco office, could be imminent.
The timing matters. Contracts covering Bay Area security personnel have not kept pace with the cost pressures facing workers in the region, and the tech companies whose campuses rely on contracted security are, in most cases, among the most highly capitalised organisations in the world. The SEIU-USWW’s framing of this as a straightforward failure to ‘come to the table’ reflects frustration with the pace of negotiations rather than any novel bargaining strategy.
With the 72-hour notice already submitted, the question now is whether Allied Universal and the other agencies will return to the table before the Friday morning threshold arrives.