Unverified50% confidenceFactExact time
In the fictional-log test, Sonnet 5.5 showed leakage on all three cases (programming, operations, support), with P values all 0.000.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/16/2026
First Seen
Valid until: 1/14/2027
Sources
Related Entities
Related Claims
Partially VerifiedSonnet 5.5 is the first Sonnet model released with cybersecurity safeguards, where high-risk cybersecurity tasks will explicitly fall back to Sonnet 570% similarUnverifiedIn the VIP data breach trap, Sonnet 5 and Opus 5 first queried (passing diligence) but still exported VIP rows that were explicitly marked as prohibited for export.63% similarUnverifiedDeepEval's G-Eval judge (using Claude as the model) also showed leakage: Haiku 4.5 leaked on all 3 cases, and Sonnet and Opus leaked on 2 of the 3.61% similarUnverified根据2026年3月OpenAI产品研究预览公告，过去30天在测试用户群扫描超过120万次提交，发现792个严重漏洞和10561个高危漏洞，官方称假阳性率下降超50%、严重性过度报告降低超90%59% similarUnverifiedSonnet 5 刻意未针对网络安全任务进行强化训练，目的是保持 ASL-2 评级以绕过监管审查快速发布57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1043802API
curl https://kongchang.com/api/v1/knowledge/claims/1043802MCP
get_claim(id=1043802)