Unverified50% confidenceFactExact time
OpenAI published a blog post stating that protections against adversarial distillation have been applied to GPT 6.2 Astra, with the attack success rate in internal red-team testing dropping from 5% to 0.2%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/21/2026
First Seen
Valid until: 2/19/2027
Sources
Related Entities
Related Claims
UnverifiedOpenAI's head of safety systems stated that GPT-6.1 Astra showed improvement on some tasks but fell short of standards in 'operating within authorized boundaries' and 'accurately reporting completed work.'72% similarUnverifiedOne adopted GPT-5.6 Luna fix reduced attack violations from 10.7% to 0% and raised safe task completion from 89.7% to 100%.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1193879API
curl https://kongchang.com/api/v1/knowledge/claims/1193879MCP
get_claim(id=1193879)