Partially Verified65% confidenceBenchmarkExact time
2024 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%
3
Sources
65%
Confidence
Long-term
Relevance
8/8/2026
First Seen
Sources
Related Claims
UnverifiedPrevious industry assessments of GPT-4's cybersecurity capabilities rated them as between 'medium' and 'high'.75% similarUnverified2023年多项学术研究证明GPT-4、Claude等主流大模型均对提示注入攻击存在不同程度的脆弱性73% similarUnverified2024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%71% similarUnverifiedGPT-4在2023年初于MMLU上取得86.4%的成绩69% similarUnverifiedModels such as GPT-4 demonstrated vulnerability discovery and exploitation capabilities in CTF competitions during 2022–2023.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/707501API
curl https://kongchang.com/api/v1/knowledge/claims/707501MCP
get_claim(id=707501)