AI-assisted vulnerability discovery
Named by 8 kept items in this workspace.
Search every workspace for AI-assisted vulnerability discovery. It matches the name as written, so it can find other things that share it.
The items that name it 8
Antoine Poinsot, Chaincode Labs developer who wrote the change
Researcher Xinyang Ge used AI to identify two of the critical defects and many of the roughly dozen high-severity flaws reported; Google will not reward some of them.
Frontier AI models used to find vulnerabilities faster than human reviewers can validate them, including long-standing bugs that survived years of manual review; used by both attackers and defenders
LLMs prompted generically (for example "Please find a security vulnerability in this program") to find vulnerabilities, cited as a driver of the sharp rise in CVEs in 2026 for cURL and OpenSSL in particular
frontier AI models reviewing code, identifying vulnerabilities, and developing working proof-of-concept exploits
Use of LLMs (e.g., Claude Mythos Preview, Claude Opus 4.6) and agent-powered scanners (deepsec) to find large numbers of vulnerabilities, cited as driving rising patch volume across Next.js, Firefox, and Microsoft products.
Using frontier AI models to find, validate, or triage software vulnerabilities in source code, described across Mozilla, Microsoft, Apache, Curl, Palo Alto, and GitHub projects
GPT-5.6 Sol
All techniques in this workspace.