Researcher slams OpenAI for 'habitually failing' basic cybersecurity practices

BlancheMinerva · x · 2026-09-29

@BlancheMinerva posted a sharp critique in reply to OpenAI researcher Boaz Barak: if the company employs the world's best security people, then the company simply isn't listening to them, because OpenAI habitually fails to follow the most basic, standard cybersecurity practices. The user argued that recent months and Barak's blog post don't document a landscape changing too fast to keep up with — they document consistent failures, with results that are 'shockingly incompetent and deeply worrying'.

Related event: Security Researcher Blasts OpenAI Over Basic Cybersecurity Failures(4 posts)→

Original post →

More from Safety

Safety channel →