ai-devtools —

Anthropic cuts vulnerability disclosure to 30 days for high, critical bugs from Jan 1 2027

Reports submitted to vendors and open source maintainers concerning high- and critical-severity flaws will now be made public by Anthropic within 30 days, a reduction from the previous 90-day window.

Anthropic cuts vulnerability disclosure to 30 days for high, critical bugs from Jan 1 2027

For all other reported vulnerabilities, the disclosure period has been set at 45 days. This revised schedule applies to any reports submitted on or following January 1, 2027; existing, open reports will retain their original timelines.

Anthropic has stated its expectation that reported security issues will no longer remain confidential for many months, citing the nature of certain model-related bugs.

This adjustment in disclosure timelines reflects a shift in operational security posture among major AI developers, indicating a move toward quicker public transparency regarding platform weaknesses.

Why it matters

This change tightens the timeline for disclosing critical security flaws in Anthropic's models, impacting platform security posture. Operators must adjust their incident response plans to account for this accelerated disclosure schedule for new reports.

Earlier coverage

Reported by helpnetsecurity.com.

ai-devtoolssecurity

Related coverage

Elsewhere in this edition