Verified, then logged.
Sun 22 Nov 2026 · 00:33 KST / 15:33 UTC한국어 | ENSearch /

04 / Policy · Policy

Anthropic cuts vulnerability disclosure window to 30 days

DigitalToday reports that the policy revision reflects the growing likelihood of AI repeatedly finding the same vulnerability.

Key points

  1. 01DigitalToday's 17 Nov 2026 report on Anthropic's CVD policy revision
  2. 02Software vulnerability disclosure, with High and Critical severity classifications cited
  3. 03Disclosure window shortened from a maximum of 90 days to 30 days; application from 1 Jan 2027

DigitalToday, a Korean technology news outlet, published a report on Anthropic's revised coordinated vulnerability disclosure (CVD) policy on 17 Nov 2026. According to the report, Anthropic, an AI company, announced that the new criteria apply from 1 Jan 2027.

Scope. DigitalToday reports that the revised policy covers the period before software vulnerabilities are publicly disclosed. The report refers to vulnerabilities classified as High or Critical severity in connection with the new policy.

Disclosure period. According to DigitalToday, Anthropic is shortening the vulnerability disclosure window from a maximum of 90 days to 30 days.

Reason for the change. DigitalToday attributes the revision to rapid advances in AI's ability to detect security vulnerabilities. The growing likelihood that AI will repeatedly discover the same vulnerability prompted a review of existing disclosure practices.

The next step reported by DigitalToday is the application of the new criteria on 1 Jan 2027; the report does not specify whether exception conditions have been finalised.