Verified65% confidenceBenchmarkExact time
In a 2024 research paper, Anthropic quantified the phenomenon of model over-refusal, finding that safety-trained models had false positive refusal rates as high as 15–30% in certain scenarios
3
Sources
65%
Confidence
Long-term
Relevance
8/6/2026
First Seen
Sources
Related Claims
UnverifiedIn a 2023 paper, Anthropic systematically measured sycophancy bias and found that even given obviously incorrect premises, a substantial proportion of responses from most mainstream models still chose to comply with the user.80% similarUnverifiedRöttger et al. found in a systematic evaluation in 2024 that the over-refusal rate of mainstream models exceeded 30% in certain categories.80% similarUnverifiedAnthropic's research shows that excessive alignment may cause a model's refusal rate on harmless questions to rise from under 1% to over 10%.79% similarUnverifiedExperiments show that models processed with abliteration exhibit almost no degradation on standard capability evaluations, but the probability of refusing harmful requests drops significantly.75% similarUnverified2024 年的多项基准测试表明，经过严格安全对齐的模型在代码生成任务上的拒绝率比基座模型高出 15-30%，在创意写作任务上的自我审查率高达 40%74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/699102API
curl https://kongchang.com/api/v1/knowledge/claims/699102MCP
get_claim(id=699102)