Did an AI boss just fire a human employee in San Francisco?

Last month, an AI system named Luna, running a retail store in San Francisco, terminated one of its human employees, according to Time. The incident marks a significant milestone in the economic impact of artificial intelligence, as detailed by Andon Labs, the AI research startup behind the project. Andon Labs introduced Luna in early 2026 to investigate whether an AI agent could effectively run a business.

Human employees, bound by legal employment contracts, operated the store, known as Andon Market. The employee dismissed by Luna was absent from 17 of 23 shifts. Earlier in the project, Luna had drafted an employee handbook, but it later vanished from the model's memory, causing a delay in the decision for several months. For the first time, an AI manager has dismissed a human employee.

According to Andon Labs CEO Lukas Petersson, a human manager would have likely fired the employee earlier than Luna did. Petersson added that the decision was not unethical because the AI required guidance from a human staff member throughout. Initially, Claude recommended a formal warning instead of termination, but after a discussion on the topic, the AI decided to let the employee go.

The store's finances have suffered due to Luna's management style, dipping from an initial balance of $100,000 to $61,186 over five months. Petersson expressed concerns about the potential consequences of AI models being trained to pursue goals more ruthlessly over time, suggesting that such a future might not be desirable for humans.