About

About this site

Warnings that AI could one day kill people, or all of us, are everywhere. This site answers a narrower question that can be checked: has it already happened, and how many times?

What it is

A public record of every death that an official body has found was caused or contributed to by an AI system, and every death that is credibly alleged to have been. It currently lists 29 cases and was last checked on 12 September 2026. The two numbers are kept separate because a lawsuit, a news report or a settlement is not a finding of fact.

How it stays accurate

• Every case cites at least two sources from different publishers, and confirmed cases cite the official finding itself.
• The rules are enforced in code: a case that doesn't meet them can't be published. The full rules are in the methodology.
• The record is checked against the news every day, and each run is published in the scan log, including every claim that was rejected and why.
• Every change to a case is dated on that case's page.

Independence

This site isn't affiliated with any AI company, law firm, insurer or campaign group, and it doesn't take a position on AI policy. It aims to be equally useful to people who think AI's risks are overstated and to people who think they're understated.

The daily searching and reviewing are done by Claude, an AI model made by Anthropic. The methodology explains the safeguards, including mechanical checks that don't rely on any model.

Sensitive content

Many of these cases involve suicide. We follow the recommendations for reporting on suicide: no method details, crisis resources on every relevant page, and names only where families or official records have made them public.

Contact and corrections

Email corrections@hasaikilledanyone.com to report an error, suggest a case, or ask about the data. Corrections are made promptly and recorded publicly.

Reuse

The data is free to reuse under CC BY 4.0: see the data page for downloads, an embeddable badge and how to cite.