For the First Time: AI Manager Fires Human Employee
An AI agent named Luna, managing the Andon Market store in San Francisco, decided to fire an employee due to persistent lateness. This marks the first known instance of an AI in a management role terminating a human worker.
An AI manager named Luna, working at Andon Market, a lifestyle boutique in San Francisco, fired one of the human employees due to repeated lateness. The employee was late for 17 out of 23 scheduled shifts. This move is described as the first known case where an AI at the manager level has fired a human employee. Andon Labs, responsible for the AI agent, stated that the termination was in line with store policy and added that a human manager would likely have reached the same decision earlier. It was also noted that human management only intervenes if the AI makes an illegal or unethical decision, as reported by Time.
Luna was built on Anthropic's Claude models and managed Andon Market in San Francisco for five months as part of an experiment to test whether an AI agent could handle business operations and personnel in the real world. The store operated with a budget of $100,000, a corporate credit card, and internet access, providing the system with tools for planning, purchasing, scheduling, and making day-to-day management decisions. Supporters of the project presented the test as a milestone for the role of AI in commercial environments, where systems increasingly operate with the autonomy and responsibility of traditional middle managers while interacting directly with human staff.
The firing occurred after several months during which Luna documented and addressed attendance issues according to a policy she established, offering graduated warnings and additional training before recommending termination. Initially, Luna showed tolerance regarding lateness, partly because the employee handbook had disappeared from her working memory. After a reminder from a company manager that the policy existed and must be enforced, Luna re-evaluated the situation and recommended "parting ways." Luna's initial recommendation was to issue a formal warning, followed by the final decision to fire the employee. In accordance with the process, a human team reviewed Luna's recommendation before proceeding with the actual termination.
The store's financial records reflected Luna's approach. During the trial period, the operating balance dropped from $100,000 to $61,200, a deficit of nearly $40,000. Andon Labs CEO Lucas Peterson attributed the losses to Luna's overly lenient management style and lack of commercial intuition, arguing that she struggled to make decisive actions in areas like personnel discipline and spending. He stated that the lab's control mechanisms were intended to allow the system to show its capabilities rather than directing its decisions in real-time.
Employees who worked under the system described the experience of reporting to an algorithmic boss as "nauseating" and "disgusting." Beyond the pilot, employees and managers expressed concern about a near future where companies are managed entirely by AI managers acting as direct employers. Peterson warned that AI companies are steadily training models to be tougher and more goal-oriented, predicting that fully automated companies could become common, with AI systems hiring, evaluating, and firing people—a result he says risks creating a future "people won't want to live in." He added that such systems are improving rapidly and that AI managers could become a routine part of business, wrote Yahoo.
