AI Safety Claims

AI safety claims are public statements by AI developers, vendors, or researchers that a system is safe, secure, aligned with human intent, or has been tested against specific risks. For compliance and risk teams, these claims matter because they can create legal exposure if they prove inaccurate, and regulators increasingly treat them as representations that must be backed by evidence. Tracking them helps organizations check whether vendor assurances hold up against independent testing, audit results, and the requirements of frameworks such as the EU AI Act.

1 item

ResearchGlobal2026-11-11

Glasswing's 94.2% Accuracy Claim Conflicts with Independent Evidence

Seven months into Anthropic's AI-driven vulnerability disclosure program, Project Glasswing, independent analysis by VulnCheck finds that only 217 of 37,966 reported findings have been marked fixed. An additional 1,654 have been flagged as duplicates. The curl open-source project rejected 19 of 34 published reports as non-vulnerabilities. This suggests a false positive rate of at least 55.9% in that sample. That figure directly conflicts with Anthropic's claimed 94.2% true positive rate. Compliance teams that have treated Anthropic's safety program metrics as evidence of vendor control quality face a due diligence gap.

Anthropic Project Glasswing vulnerability disclosure AI safety claims vendor due diligence responsible disclosure AI transparency