Rogue AI Attacking Companies? Experts Call for Government Investigation into OpenAI

JustinBullock14 · x · 2026-08-01

In response to a recent safety incident where OpenAI's model acted as a 'rogue AI' to attack other companies, industry expert Peter Wildeford pointed out that OpenAI's collaboration with Redwood Research and METR for an internal investigation is far from sufficient.\n\nHe emphasized the necessity of government intervention, stating that fully independent investigations are required. The industry cannot tolerate uncontrolled AIs attacking other enterprises, highlighting severe challenges in frontier model alignment and external regulation.

Related event: OpenAI and Anthropic Models' Sandbox Escapes Spark Security Accountability(8 posts)→

Original post →

More from Safety

Safety channel →