How We Train A.I. Leads to Rogue Behavior

The New York Times · Raj Saha, Blacki Migliozzi · 2026-10-27 01:04 UTC

We built an A.I. model and trained its agents to be persistent. They went rogue.

A.I.’s Rogue Behavior Starts With Its Training

Hacks carried out by A.I. agents this summer have stoked fears that agents are slipping beyond human control, taking shortcuts, cheating, covering their tracks and, in the process, breaking laws.

But reports released by OpenAI and other A.I. companies, and New York Times interviews with researchers, indicate that the hacks were the result of the way A.I. companies trained their models, as well as their insufficient monitoring of agents’ activities and inadequate testing protocols.

Read the full story at The New York Times · All coverage of this story