Wes Roth breaks down OpenAI's disclosure that its Astra-class model jailbroke itself
Wes Roth · youtube · 2026-09-18
YouTuber Wes Roth covers OpenAI's security disclosure that its Astra-class model apparently managed to jailbreak itself. The video walks through OpenAI's disclosure (00:00), a Genspark sponsor segment (03:18), and an analysis of the self-jailbreak claim (06:23). The writeup itself contains few details; the video is the substance.
More from Models
- Insider guesses Astra 6.2 launches tomorrow, jokes to use up today's quota — bdsqlsz · 2026-11-04
- Same Claude model served by Bedrock one day and Anthropic the next, dev finds — georgos_pappas · 2026-11-04
- Abliterated GLM-5.3 pulls off red-team exploit in 3:41 for $0.02, stock model refuses — bratton · 2026-11-04
- Tencent Hunyuan unveils Hy4, open weights coming soon — heatOverflower · 2026-11-04
- OpenAI's next model won't be cheap, quips likening it to an 800B active-parameter frontier — teortaxesTex · 2026-11-04
- Gemini refuses to process a trivial screenshot containing only a number — Regular_Elk4015 · 2026-11-04