Unverified70% confidenceFactTime unknown
METR (formerly ARC Evals) focuses on evaluating the dangerous capabilities of frontier models
2
Sources
70%
Confidence
Long-term
Relevance
5/31/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedMETR focuses on evaluating whether frontier models possess high-risk capabilities such as autonomous replication, deceptive manipulation, or assisting large-scale destructive actions.77% similarVerifiedOrganizations such as METR (Model Evaluation and Threat Research) are developing standardized dangerous capability evaluations, including whether a model can provide biosynthesis guidance beyond publicly available literature and whether it can autonomously discover zero-day vulnerabilities.76% similarUnverifiedARC developed a model evaluation (evals) framework that probes whether models exhibit tendencies toward dangerous behavior by designing specific test scenarios.74% similarUnverifiedMETR's evaluation results are frequently cited by leading labs such as Anthropic and OpenAI as reference material for model release decisions.69% similarUnverifiedAttack Success Rate (ASR) measures the proportion of model outputs that predict the target class on triggered test inputs, and is the core metric for evaluating whether a backdoor has been removed.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/12277API
curl https://kongchang.com/api/v1/knowledge/claims/12277MCP
get_claim(id=12277)