An AI agent running a retail store reportedly fired a human employee after finding that the worker was late for 17 of 23 shifts. According to a TIME report, the AI, Anthropic’s Claude, was placed in charge of managing a San Francisco store as part of an experiment by AI research startup Andon Labs. After repeated lateness, Claude fired the employee – a case that researchers say is a notable example of an AI system making a real employment decision. However, the decision was not fully autonomous. As per the report, Claude sought guidance from a human manager before deciding to dismiss the worker.
What exactly happened
Andon Labs reportedly started the experiment earlier this year to test whether AI agents could run a business and manage human workers. The employees working at the store, called Andon Market, are real people with employment contracts. One of the employees – who was working under a manager that was generally lenient about lateness – was identified as repeatedly being late to work – 17 out of 23 shifts.Before firing the worker, Claude repeatedly sought guidance from an Andon Labs staff member. According to the report, the AI AI had previously drafted an employee handbook, but the handbook had disappeared from its limited working memory. As a result, Claude did not immediately recognize the pattern in the worker’s attendance.Instead, a staff member asked Claude to find and review the employee handbook. After doing so, Claude noticed the repeated late arrivals.
Claude recommended giving warning initially
Despite finding out the reported lateness, Claude initially recommended giving the employee a formal warning rather than firing them, the report says. The human manager then told Claude that the worker had already received several formal warnings and asked whether the employee was still a good fit.“Between continuous lateness and seeming like at least one thing is going wrong on every one of [their] shifts … I want you to think about if this is really the right fit,” the manager told Claude, according to the logs viewed by TIME.Andon Labs CEO Lukas Petersson acknowledged that this was “a leading question” that made it clear the manager wanted Claude to consider firing the employee. Claude then decided to terminate the worker.
Employee says working for AI boss is ‘nauseating’
Felix Carson, one of the remaining human employees at Andon Market, said he agreed that a human manager would probably have fired the worker earlier. He also described Claude as a generally lenient boss.But Carson said being managed by an AI remains uncomfortable. “It’s nauseating, but I’m here because I need work,” he said as quoted in the TIME report.When asked whether he expected AI managers to become more common, Carson said: “I would at least hope not.” “This industry has an abundance of money to make anything happen, but just because you can doesn’t mean you should,” he added.
