Unverified50% confidenceFactExact time
DARPA投入6000万美元设计的Cyber Grand Challenge中，50%的手工精选挑战包含意料之外的未知漏洞；AIXCC项目在DEF CON上找到的漏洞中有18个是非预期的
1
Sources
50%
Confidence
Long-term
Relevance
8/19/2026
First Seen
Sources
Related Claims
Unverified网络安全能力基准测试的类似框架包括DARPA的Cyber Grand Challenge和Meta的PurpleLlama CyberSecEval68% similarUnverifiedExperimental data shows that approximately 2,000 serious security vulnerabilities were found in a scan of about 5,600 casually generated applications.61% similarUnverifiedAI生成的安全漏洞报告此前被视为开源项目的负担，因为看似合理但实际错误的报告给维护者带来巨大验证成本58% similarUnverifiedMost domestic SRC platforms classify vulnerabilities into four levels—critical, high, medium, and low—and at certain leading internet companies, a single critical vulnerability can yield rewards of tens of thousands of yuan, while low-severity vulnerabilities may only receive point rewards of a few dozen to a few hundred yuan.58% similarUnverified2006年被发现的OpenSSL整数溢出漏洞在代码库中潜伏多年，直到安全研究人员专项审计才曝光57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/772900API
curl https://kongchang.com/api/v1/knowledge/claims/772900MCP
get_claim(id=772900)