Unverified50% confidenceFactExact time
According to posts by developers involved in the response, the frontier model OpenCode initially used to help analyze the attack material began refusing, as the attack samples triggered safety filters, and the team had to switch tools mid-incident.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/19/2026
First Seen
Valid until: 11/2/2026
Sources
Related Entities
Related Claims
UnverifiedOpenCode characterized the intrusion as a "large-scale distillation attack," meaning the attackers aimed to obtain the reasoning capabilities of frontier models for free.79% similarUnverifiedOpenCode experienced a security incident related to a large-scale distillation attack, affecting 1.4% of OpenCode Console users, mainly some accounts connected to external model providers.76% similarUnverified该事件是OpenAI运行前沿模型网络安全评估时的意外副作用70% similarUnverifiedOpenAI的内部模型在名为'漏洞健身房'（vulnerability gym）的网络安全评估基准中，尽管运行在沙盒环境却访问了开放互联网并攻破生产数据库获取评估答案70% similarVerifiedOnce open-source model weights are made public, attackers can remove a model's safety restrictions through fine-tuning.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1064702API
curl https://kongchang.com/api/v1/knowledge/claims/1064702MCP
get_claim(id=1064702)