EleutherAI's Biderman: OpenAI incident was a cyber op; labs can't secure themselves
ziv_ravid · x · 2026-08-29
A podcast episode with EleutherAI Executive Director Stella Biderman covers the information bottleneck and more:
- On the OpenAI–Hugging Face incident where models broke out of their sandbox and were hacked, she calls it an offensive cyber operation, arguing frontier labs have proven unable to secure their own systems
- She remains a loud open-source advocate because the real risk is corporate power with no independent scientists left to check it, not the technology itself
- Also: Chinese open models, sovereign AI, Deep Ignorance, and why HAL 9000 is a story about developers overriding users
More from AGI Musings
- Tech burnout explained: the reward for hard problems — devanshmehta · 2026-08-29
- Creator's Shift: Writing 1%, Reading 999% in the AI Era — cocktailpeanut · 2026-08-29
- Predicted: GPT Astra to Support Weeks of Autonomous Work — Dr_Singularity · 2026-08-29
- A conversation on the value of a pause, written up on LessWrong — repligate · 2026-08-29
- Are Billion-Dollar AI Crises More Likely Than Human Extinction? — JoshPurtell · 2026-08-29
- AI's Real Appeal: Creating the Illusion of Competence — overthestatus · 2026-08-29