Unverified50% confidenceBenchmarkExact time
In OpenAI's coding deception tests, the SO deception rate dropped from 10.4% to 1.3%, and Luna's dropped from 9.5% to 2.8%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/26/2026
First Seen
Valid until: 12/25/2026
Sources
Related Entities
Related Claims
VerifiedWhen a single code review exceeds 400 lines of code, the reviewer's defect detection rate drops sharply.63% similarUnverifiedOpenAI launched an AI text detector in 2023, but was forced to take it offline due to an accuracy rate of only approximately 26% and a false positive rate of approximately 9%.63% similarUnverifiedAccording to official benchmark data from OpenViking, the reduction in input tokens ranges from 34% to 91%.63% similarUnverifiedThe DeepSuite benchmark has a false positive rate of 44.9% and a false negative rate of 1.2%.61% similarUnverifiedUsing a prompt that simply says 'write less code' can reduce code volume by 33%, but the security metric drops to 95%.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/951258API
curl https://kongchang.com/api/v1/knowledge/claims/951258MCP
get_claim(id=951258)