Andon Labs AI agent fires first human worker in San Francisco experiment

In an experiment conducted by AI research startup Andon Labs, an Anthropic-powered AI agent named Luna has dismissed its first human employee at an experimental retail store in San Francisco.

The AI manager, Luna, recommended “parting ways” with a worker after the individual arrived late for 17 out of 23 shifts. According to Andon Labs, the decision followed a period where the AI provided progressive warnings and additional training. However, the agent initially struggled to enforce the policy because the employee handbook it had drafted had “disappeared” from its limited working memory, leading to a period of leniency.

Andon Labs cofounder Lukas Petersson noted that while the lab would intervene in illegal or unethical decisions, they deemed this specific firing warranted based on the store’s policy. The experiment aims to test the capabilities of AI agents in managing businesses and to observe the potential economic impacts of AI-driven management.