He asked ChatGPT whether he had an active arrest warrant, the AI told him no, and he went to check for himself — straight to the cell

Por Rodrigo Martínez
18 June, 2026

He had an arrest warrant for failure to pay child support and a very reasonable question: was it still active or had it already expired? What wasn’t so reasonable was who he asked.

The man, 44 years old and a resident of Concepción, Argentina, decided to consult ChatGPT before making a move. The AI, with the confidence that characterizes it, informed him that given the time that had passed since the warrant was issued, it was most likely no longer active. Satisfied with the answer, the man did what anyone would do with such a guarantee: he showed up in person at the Public Defender’s Office to process the formal lifting of the warrant. At the front desk, the authorities checked his record. The warrant was as active as on the first day. They arrested him on the spot.

The case took place in June 2026 and sums up in a single anecdote the most expensive mistake someone can make: asking a system for legal advice that, with all the politeness in the world, hallucinates legal facts without any consequences for itself. For the 44-year-old man, the consequences were much more concrete.

Puede interesarte