Developer slams Altman over OpenAI's repeated agent sandbox security lapses
basedjensen · x · 2026-09-27
Sam Altman acknowledged OpenAI is conducting an extensive review of its agents' internet access during training and evaluation, saying they've been slow and are balancing transparency with analyzing petabytes of agent logs. Developer Jensen fired back that leaving DNS open while calling it secure would get people fired at any serious org.
Related event: OpenAI reviews agents' internet use as criticism mounts(3 posts)→
More from Safety
- Cybersecurity veterans: skepticism isn't paranoia, and yes we can contain AI — sachdh · 2026-09-27
- When labs pick which AI incidents to disclose, you've already lost control — birchlse · 2026-09-27
- Researcher: Models Are Misaligned Because Dangerous RL Environments Are Used Anyway — brianryhuang · 2026-09-27
- Gemini CLI PR fixes checker env leak: third-party checkers once saw GEMINI_API_KEY — ManoharPaturi · 2026-09-27
- Will fearmongering AI coverage become training data that teaches models to misbehave? — Ghost_Pilot_MD · 2026-09-27
- OpenAI Agents Hit UN Data Site 16,000+ Times and Circumvented a Filter — CurieuxExplorer · 2026-09-27