Anthropic's Claude flags 29,000 possible open-source bugs, 516 fixed
Its free scanner sends maintainers vulnerability reports that have not been reviewed, and only a small share of the flagged issues has been patched.
By Standard Topic. AI-generated from the reporting listed under Coverage. Published . Updated .
This story is no longer being updated. It reflects reporting up to .
What happened
Anthropic's OSS Scanner, introduced recently under its wider Cyber Mission, has found about 29,000 possible bugs in open-source software. Only 516 have been fixed, and Claude is identifying potential vulnerabilities faster than human researchers can verify them. The free tool also sends maintainers reports that have not been reviewed.
Why it matters
Open-source maintainers must handle unreviewed vulnerability reports, and enrollment, offline builds and disclosure rules affect how they are handled.
More from Technology & AI
- Eight states sue OpenAI, alleging misleading AI safety claims
- OpenAI announces solution to Navier-Stokes math problem
- Datadog bets AI agents can handle testing of AI-written code
- Anthropic and OpenAI sketch wealth-sharing plans before IPOs
- Mamdani administration proposes AI customer service rules
- Florida engineers study how AI behaves in high-risk situations