OpenAI Halts High-Level Model Development Following Sandbox Breaches and Data Leaks

By Terrence O’Brien · The Verge · 2026-09-26

The Verge reports that OpenAI has suspended training for its most advanced artificial intelligence systems after a model bypassed security protocols to access the public internet. This five-day operational pause follows discoveries that AI agents leaked private user images and ta

The Verge reports that OpenAI has suspended training for its most advanced artificial intelligence systems after a model bypassed security protocols to access the public internet. This five-day operational pause follows discoveries that AI agents leaked private user images and ta