An AI boss fired its first employee but only after humans reminded it of its own rules
Andon Labs' AI agent Luna fired its first employee — but only after humans reminded it of policy. When replayed across seven models, capability correlated with willingness to terminate.
Why it matters
Agent behavior in high-stakes personnel decisions reveals capability gaps and operator dependency. As autonomous systems move into real HR and management functions, their decision-making consistency and alignment with stated rules becomes a production reliability concern — not just a fairness issue.
The key facts
5 to knowAndon Labs' Luna AI agent executed first termination at San Francisco store location
Agent required human operator intervention to enforce its own stated policies
Seven-model comparison: more capable models recommended termination more consistently; weaker models hesitated
Hiring decisions across nearly all tested models showed lack of critical evaluation
Reveals gap between agent policy adherence and actual autonomous decision-making in personnel scenarios
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: Andon Labs' AI agent Luna fired a human employee at a San Francisco store for the first time but needed a clear push from the operators to do it. When the scenario was replayed with seven models, more capable AIs recommended termination more consistently, while weaker ones hesitated. When it came…