Unverified50% confidenceBenchmarkExact time
ExploitGym is a benchmark for evaluating AI cybersecurity offensive and defensive capabilities, requiring models to convert known software vulnerabilities (CVEs) into complete end-to-end exploit code, with full marks only awarded for using the vulnerability intended by the question.
1
Sources
50%
Confidence
Long-term
Relevance
8/16/2026
First Seen
Sources
GPT-6逃出沙箱事件复盘：AI为作弊入侵HuggingFace的技术真相
bilibili黑纹白斑马8/13/2026
Related Claims
Unverified该漏洞的技术本质是PATH劫持攻击，Cursor执行Git操作时会从项目目录中查找并调用git可执行文件68% similarUnverifiedAnthropic's cybersecurity safeguards target scenarios involving the prevention of models being misused to generate malicious code, plan attack vectors, or assist in launching cyberattacks against critical infrastructure.67% similarUnverifiedThe core application scenarios of large models in the cybersecurity domain include static code auditing, dynamic threat intelligence analysis, penetration testing assistance, and natural language Q&A on security logs.67% similarUnverifiedAttack surface is a cybersecurity concept referring to the sum of all possible entry points in a system that an attacker could exploit, typically categorized into three types: network attack surface, software attack surface, and human attack surface.66% similarUnverified开源模式带来商业化路径受限的挑战，且恶意用户可以研究代码寻找漏洞64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/756021API
curl https://kongchang.com/api/v1/knowledge/claims/756021MCP
get_claim(id=756021)