Blanche Minerva: OpenAI leadership has ignored computer security basics for years
BlancheMinerva · x · 2026-09-30
Former OpenAI safety researcher Blanche Minerva clarified in a debate with Boaz Barak and others that she is not attacking OpenAI's security engineers, but the corporation: leadership has long ignored the security team, which lacks real power to force change. She argues computer security alone doesn't solve AI safety, and OpenAI's weak security practices leave no basis to assess whether they could prevent hacks carried out by its models.
More from Companies & People
- Lisa Su's Week: $8.2B World Labs Deal, AMD First Woman-Led $1T Company — alliekmiller · 2026-09-30
- Modal's Runtime conference lands Oct 1 in SF with Cognition, Anthropic, NVIDIA speakers — charles_irl · 2026-09-30
- UChicago Booth hires Applied AI faculty: IC track, department-funded research — ethayarajh · 2026-09-30
- Yacine recalls interviewing Rajan Nagarwal amid 30-second WiFi drops and a fire alarm — yacinelearning · 2026-09-30
- A masterclass on the one-person AI creative team: two marketers push AI to its limits — alexgoughcooper · 2026-09-30
- GPT-6 naming chaos: Terra became Sol, Sol became Astra, out came GPT-6.1 Sol — Balance- · 2026-09-30