Unverified50% confidenceBenchmarkExact time
In the AgentDyn benchmark, environmental guidance reduced the attack success rate to 0%.
1
Sources
50%
Confidence
Long-term
Relevance
9/30/2026
First Seen
Sources
Related Entities
Related Claims
Unverified卡内基梅隆大学研究显示对抗性图案攻击成功率在受控环境为90%以上，但引入正常视角变化后降至实际部署场景的30%以下70% similarUnverifiedResearch indicates that an attacker needs to modify as little as 0.01% of training data to potentially implant backdoor behaviors in a model that are undetectable by standard evaluations.69% similarUnverifiedAnthropic conducted red-team testing on 123 use cases using an agent, finding an attack success rate of over 20% in autonomous mode, which dropped to 10% after adding mitigations, and can be reduced to below 0.1% with the latest protective mechanisms.69% similarUnverifiedThe author deliberately introduced a genuine severe regression that reduced an agent's success rate from 92% to 52%, yet the statistical rule originally designed for this purpose passed the PR because the regression was concentrated in a small number of tasks rather than being uniformly distributed.66% similarUnverifiedIn real-environment A/B testing, the post-trained checkpoint reduced the tool-call failure rate by 21.2% relatively compared to the earlier version.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/961205API
curl https://kongchang.com/api/v1/knowledge/claims/961205MCP
get_claim(id=961205)