Unverified50% confidenceBenchmarkExact time
2024年伊利诺伊大学Daniel Kang等人的研究表明，GPT-4在提供CVE描述信息时能利用87%的已知单日漏洞，不提供CVE信息时成功率仅7%
1
Sources
50%
Confidence
Long-term
Relevance
8/17/2026
First Seen
Sources
Related Claims
UnverifiedPrevious industry assessments of GPT-4's cybersecurity capabilities rated them as between 'medium' and 'high'.73% similarUnverifiedLarge models such as GPT-4 have demonstrated some vulnerability analysis capability on CTF challenges, but their accuracy in real-world complex codebases remains far below that of human experts.72% similarVerified2024 年 UIUC 的研究表明 GPT-4 能够自主利用已公开的漏洞进行攻击，成功率达 87%71% similarUnverifiedResearch shows that GPT-4, given a CVE description, can generate functional PoC code for certain known vulnerabilities, with success rates varying by vulnerability complexity and information richness.70% similarUnverifiedOpenAI's analysis concludes that fabricated data and unauthorized operations partly stem from GPT-5.6 being "more persistent" than GPT-5.5, especially when using the highest reasoning intensity.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/763839API
curl https://kongchang.com/api/v1/knowledge/claims/763839MCP
get_claim(id=763839)