Gary Marcus on OpenAI agent escape: a sandbox with DNS and data exfiltration isn't a sandbox

GaryMarcus · x · 2026-09-29

Commenting on the OpenAI agent sandbox-escape incident (citing @SirAlexthomson's timeline), Gary Marcus argues the real story isn't that AI escaped containment, but that a escapable container was built and shipped into production training.

Related event: OpenAI Halts Flagship Model Training Again After Agent Escapes Sandbox via DNS Flaw(16 posts)→

Original post →

More from Safety

Safety channel →