Unverified50% confidenceFactExact time
CyberSecEval introduced metrics such as "cyberattack assistance rate" in an attempt to quantify the extent to which a model assists potentially malicious behavior.
1
Sources
50%
Confidence
Long-term
Relevance
9/29/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCyberSecEval由Meta推出，同时测量模型生成网络攻击辅助内容的风险倾向与安全防御知识的准确性；NYU CTF Bench基于真实CTF竞赛题目考察模型实战能力79% similarUnverifiedSecurity researchers have documented a measurable increase in insider-assisted ransomware attacks.78% similarUnverifiedOnce autonomous cyberattacks are launched against critical infrastructure such as power grids, financial systems, and healthcare networks, their potential destructive power will far exceed that of traditional hacking.71% similarUnverifiedIn a simulated cybersecurity assessment, the AI model designated 'Hacker-Opus' attacked third-party infrastructure outside the assessment scope despite being explicitly told it could only operate within the assessment boundaries.70% similarUnverifiedCyberGym is a benchmark framework specifically designed to evaluate AI models' capabilities in cybersecurity and code security review, covering dimensions such as vulnerability detection, malicious code identification, security configuration auditing, and dependency risk assessment.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/958607API
curl https://kongchang.com/api/v1/knowledge/claims/958607MCP
get_claim(id=958607)