Claude just fired its first human employee

Key points

• Claude was tasked with managing an Andon Market store in San Francisco.
• The AI identified an employee who was late for 17 out of 23 shifts.
• After initial leniency, Claude explicitly recommended the employee be terminated.

This is a fascinating glimpse into the shift from AI as a "tool" to AI as a "manager." We've seen AI handle data and scheduling for years, but making a definitive call on someone's livelihood is a different level of autonomy.

What's interesting here is the "personality" shift. Claude started out lenient, almost mimicking a soft-touch human manager, but eventually pivoted to a cold, data-driven conclusion based on the attendance record. It proves that while AI can be programmed for empathy, the underlying logic remains purely mathematical. If the numbers don't add up, the AI doesn't feel the social awkwardness of a firing—it just sees an inefficiency that needs to be removed. We are moving toward a world where your performance review might be conducted by a token-predicting model.

*