Unverified50% confidenceFactExact time
OpenAI stated that it will invest additional control measures to ensure that subsequent development of Astra proceeds under safe and controllable conditions.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/10/2026
First Seen
Valid until: 11/8/2026
Sources
Related Claims
UnverifiedOpenAI stated it is working to make Astra broadly available and to put its advanced cybersecurity capabilities in the hands of defenders.80% similarUnverifiedOpenAI has assessed its upcoming new model Astra as the first "Critical"-level model in the cybersecurity domain under its preparedness framework.76% similarUnverifiedOpenAI's response measures included tightening infrastructure configuration, conducting a joint forensic investigation with Hugging Face, disclosing zero-day vulnerabilities, adding Hugging Face to the Trusted Access program, and strengthening training evaluation safeguards.68% similarUnverifiedOpenAI's Agent sandbox architecture revolves around three pillars: runtime security, persistence, and orchestration.66% similarUnverifiedOpenAI的InstructGPT、Anthropic的Constitutional AI以及Meta的Llama系列都采用了类似的安全对齐流程66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/725195API
curl https://kongchang.com/api/v1/knowledge/claims/725195MCP
get_claim(id=725195)