OpenAI says Astra is the first model it has designated at the Critical cybersecurity capability threshold under its Preparedness Framework. The company says Astra can find unknown flaws and build exploits across hardened systems without step-by-step human guidance, and that stronger safeguards plus limited alpha access come with the release. Capability numbers and safeguard scores are OpenAI's own claims. Separately, Lauren shipped Dr Eggbot v0.2 with routine health checks for Grok Bots.
01
OpenAI says Astra meets the Critical cybersecurity capability threshold under its Preparedness Framework, the first model it has designated at that level. It reports ExploitBench 100%, two internal V8 zero-days under disclosure, and jailbreak refusal rising to 91.5% from 59% for GPT-5.6 Sol. Advanced cyber access starts with alpha testers, then Daybreak Blue. Figures are OpenAI's claims.
webSource preview · opens the original↗Open original →02
Lauren (@poteto) shipped Dr Eggbot v0.2.0. The Grok Bot can health-check routines for unused or expensive ones and skim chats for friction, then suggest skills, bots, or routines. Reinstall to update.
Open original →