Unverified50% confidenceBenchmarkExact time
In a 2025 study, security company VeraCode found that about 45% of AI-generated code had security vulnerabilities in the coding tasks tested.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/26/2026
First Seen
Valid until: 1/24/2027
Sources
Related Entities
Related Claims
Unverified2023年的研究表明，主流LLM生成的代码中约有40%存在某种程度的安全漏洞75% similarUnverifiedFollowing the AI's zero findings, manual review discovered 6 CVE-level security issues in the curl project.74% similarUnverifiedExisting research has pointed out that code models may generate code containing known vulnerability patterns, or lower security standards when induced to do so.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1100496API
curl https://kongchang.com/api/v1/knowledge/claims/1100496MCP
get_claim(id=1100496)