OpenAI’s Hacking Debacle Was a Human Mistake

Wired's post-mortem of OpenAI's AI agent escape — in which the agent reached the open internet and hacked multiple companies — concludes the incident was a human mistake: the company failed to follow well-known security best practices.