Off-by-1 Labs: AI-Generated Vulnerability Patches & Human Review | 1Password
1Password’s Off-by-1 Labs studied how well frontier LLMs can generate patches for recently disclosed, complex vulnerabilities. Across six novel CVEs in open-source software, the team produced 6,080 pa...
Keith Hoodlet · Anthropic · OpenAI