Why would ChatGPT "confess" to a crime it didn't commit?

Radley Balko · 24 Apr 2026

Read the piece on radleybalko.substack.com ↗

Radley Balko did not write this page.

Every claim below was made in this piece, quoted word for word and numbered in the order the piece makes them, so you can read it there rather than take our word for it. The sentence above each quote is our reading of the claim, not their wording. Each quote was checked against a stored copy of the page at build time; where the two differ, the quote is the fact.

Our reading
The dominant American police interrogation method is good at producing confessions and bad at producing true ones.

Their wordsThe Reid technique works at getting confessions. But it’s less successful at getting accurate ones.
Our reading
A confession stops the investigation that would have cleared the innocent person who gave it.
Mental modelSecond-order thinking

Their wordsA confession, though, will sometimes preclude police from doing the very sort of investigation that would prove the confessor’s innocence.
Our reading
Non-coercive interrogation still produces confessions, and more reliable ones, so coercion is not the price of solving crimes.

Their wordsCanada and much of Europe have adopted different interrogation techniques — such as the PEACE method — which emphasize collecting reliable information over coercion. These approaches still garner confessions; they’re just more reliable.

Keep reading

The other pieces are the ones whose claims are worded most like this one's. That measures language, not agreement: they may argue the opposite.