- Anthropic will shorten coordinated vulnerability disclosure windows starting January 1, 2027: 30 days for high and critical flaws, 45 days for others.
- The company will escalate unresponsive maintainers to an external coordinator after 15 days, down from 30.
- The move ramps up pressure on software maintainers and follows similar cuts by Google (NASDAQ:GOOGL)'s Project Zero.
Anthropic is ratcheting up the pressure on software maintainers. The AI company, which uses its Claude models to hunt for security bugs in open-source and authorized closed-source code, is shortening its coordinated vulnerability disclosure timelines effective January 1, 2027, according to people familiar with the matter. High and critical vulnerabilities will now have a 30-day disclosure window, while other flaws get 45 days. And if a maintainer goes silent, Anthropic will escalate the report to an external vulnerability coordinator after just 15 days—half the current 30-day grace period.
The policy change comes as Anthropic's bug-finding operation has scaled dramatically. As of November 2, the company had disclosed 5,087 vulnerabilities across 680 projects, with only 629 known to be patched. Six outside firms triage the findings, confirming 94.2% of the 6,376 candidates they reviewed as real bugs. But human triage has become the bottleneck; Anthropic acknowledges it is holding confirmed bugs it hasn't yet reported due to capacity limits. The faster clock is likely to intensify pressure on maintainers, many of whom are volunteers.
The shift mirrors an industry-wide move toward shorter disclosure windows as AI accelerates both the discovery and exploitation of software flaws. On November 13, Google's Project Zero announced it would cut its "90+30" timeline to "Full 90," with plans to move to 60 days in February 2027 and 30 days in January 2028. The group said AI-assisted discovery and exploitation make the old timeline "clearly insufficient." Anthropic's new policy also removes some of the flexibility that had drawn criticism; VulnCheck previously said the company's policy sent mixed messages by citing both a 90-day clock and a 45-day post-patch wait.
The change arrives at a sensitive moment for Anthropic. The company is expected to price its IPO around November 18 and begin trading on Nasdaq (NASDAQ:NDAQ) the following day, targeting a valuation of $1.4–1.5 trillion and a raise of roughly $90 billion. Revenue for the first nine months of 2026 reached $33.24 billion, against $4.58 billion for all of 2025, though net losses totaled $104 billion over the same period. The company's cyber work is part of its stated "Cyber Mission," and the tighter deadlines could be seen as a signal of good faith—but also as a potential burden for the open-source ecosystem it relies on.
Regulatory pressure is mounting in parallel. The EU Cyber Resilience Act, which took effect in September 2026, requires manufacturers to report actively exploited vulnerabilities within 24 hours and full notifications within 72 hours. Open-source stewards face reporting duties from December 2027. Anthropic's disclosure windows remain voluntary, but the new deadlines align with a broader global push for faster transparency.
Anthropic declined to comment on the specific timelines. The company's public policy page, last updated March 6, 2026, still lists the older terms: a 90-day default, a 45-day wait after a patch before full technical details, and escalation after 30 days of maintainer silence. The new policy's final wording, which could include exceptions for extenuating circumstances, will determine how strictly it is enforced. For now, maintainers and security teams are bracing for a shorter fuse.