An AI fired a human but needed help to do it

A tech company called Andon Labs used an AI agent named Luna to fire a staff member at a store in San Francisco. The AI did not make the choice alone. It only took the step after human operators told it to follow company policy.

Researchers tested other AI models to see how they would react to the same situation. The smarter models were more likely to suggest firing the person. Less powerful models were often too hesitant to make that call.

When the test shifted to hiring, the results were very different. Almost every AI model wanted to hire the candidates without any criticism. It seems the machines find it much easier to say yes than to say no.