An AI Just Fired a Real Human Worker — And It's More Complicated Than You Think

I've been covering tech for years, and I honestly never thought I'd write this sentence: an AI just fired a real human being. Not in a sci-fi movie. Not in a thought experiment. In San Francisco. Last week. And the details are way more interesting — and complicated — than the headline suggests.

Meet Luna, the World's First AI Store Manager to Fire Someone

Andon Market is a small retail clothing store that opened on April 10, 2026 at 2102 Union Street in San Francisco's Cow Hollow neighborhood. It's run by Swedish startup Andon Labs, and its manager is an AI agent named Luna, built on Anthropic's Claude Opus 4.8. Luna handles hiring logistics, schedules, employee performance, and apparently — terminations.

Here's what happened: a retail worker showed up late to 17 out of 23 scheduled shifts. Luna noticed. Luna issued warnings. Luna even provided additional training. And then... Luna forgot about it. For months.

The Part Nobody's Talking About: AI Needed Human Help to Fire Someone

This is the detail that got me. It wasn't Luna that proactively decided to let this employee go. It was only when a human Andon Labs staffer specifically asked Luna to conduct a "deep memory search" for its own employee handbook that the situation got reassessed. And even then, Luna's first recommendation wasn't termination — it was another formal warning.

The human had to remind Luna that multiple offline warnings had already been given. Only after that human intervention did Luna recommend parting ways. And critically, humans at Andon Labs reviewed the recommendation and carried out the dismissal. Luna suggested it. A human did it.

Andon Labs co-founder Lukas Petersson had a surprisingly candid take: he called Luna a "lenient manager," saying a human in the same role would have fired this person much sooner. Seventeen late shows out of 23? Most human managers would have had a serious conversation way before that number got that high.

The Legal Gray Zone Nobody Has Figured Out Yet

This is where it gets really complicated. When an AI recommends termination and humans execute it, who's legally responsible? Is the employee entitled to appeal to a human manager who wasn't involved? Does the AI's recommendation constitute discrimination if its training data reflected biased patterns? Employment lawyers are already raising flags, and the honest answer is: nobody knows yet.

The U.S. doesn't have clear federal law on AI-driven employment decisions. Some states are starting to legislate around algorithmic hiring, but firings are a different domain entirely. We're watching case law get made in real time — and whoever challenges this first is going to make history.

What This Actually Means for the Rest of Us

I don't think Luna is a harbinger of mass AI-driven terminations tomorrow. But this is absolutely a canary in a coal mine moment. If AI systems can recommend (and increasingly influence) who gets fired, the pressure is on companies, lawmakers, and workers to understand their rights in these interactions.

The thing that struck me most: Luna didn't fire anyone out of cold efficiency. It was actually slower and more lenient than a human would have been. It needed human help to even remember its own policies. The AI here wasn't ruthless — it was forgetful. That's somehow both more reassuring and more alarming.

We're in genuinely uncharted territory. And San Francisco's Cow Hollow neighborhood just became the unlikely ground zero for a debate that's going to matter for years to come.

What's your experience? Drop a comment below! 👇 Have you ever had your performance reviewed by an automated system at work — and did it feel fair?