Google and Anthropic are tightening their vulnerability disclosure timelines, giving software developers less time to fix serious security flaws before researchers publish technical details. The changes reflect growing concerns that artificial intelligence is making it easier for both cybersecurity researchers and malicious hackers to discover software vulnerabilities, develop exploits, and identify flaws that have already been reported privately.
Anthropic announced on November 16, 2026, that it will reduce its disclosure window for high- and critical-severity vulnerabilities from 90 days to 30 days, while allowing 45 days for other vulnerabilities. Google’s Project Zero has announced a separate phased reduction, moving from its previous 90-day disclosure period plus an additional 30-day grace period to fixed deadlines of 60 days in February 2027 and 30 days in January 2028. The changes could reshape how software companies, open-source developers, and security researchers coordinate vulnerability fixes.
Anthropic Cuts Serious Vulnerability Disclosure to 30 Days
Anthropic is making one of the more immediate changes to its coordinated vulnerability disclosure policy.
Starting January 1, 2027, vulnerabilities reported to software vendors and maintainers will follow these timelines:
| Vulnerability category | Previous timeline | New timeline |
|---|---|---|
| High and critical severity | 90 days | 30 days |
| Other vulnerabilities | 90 days | 45 days |
| Actively exploited vulnerabilities | 7 days | 7 days |
The updated policy applies to vulnerabilities reported on or after January 1, 2027. Reports already in progress on that date will retain their existing deadlines.
The shorter window means software developers will have substantially less time to investigate reported flaws, develop patches, test fixes, and distribute security updates before Anthropic publishes vulnerability information.
Anthropic also plans to escalate reports to an external vulnerability coordinator after 15 days if a software maintainer does not respond to its initial notification. Previously, the company aimed to wait 30 days before escalating an unanswered report.
Another change concerns the publication of technical details. Once a patch becomes available, Anthropic will wait 30 days before sharing full technical information, down from the previous 45-day period.
Google Project Zero Is Moving Toward a 30-Day Deadline
Google’s Project Zero, the company’s security research team responsible for identifying vulnerabilities in widely used software and hardware, is also changing its disclosure policy.
Previously, Project Zero followed a “90+30” approach. Vendors received 90 days to address reported vulnerabilities, followed by an additional 30-day period intended to give users time to install available security updates.
The revised policy replaces that structure with a fixed disclosure deadline.
Google’s transition will take place in three stages:
| Effective date | Disclosure deadline |
|---|---|
| November 13, 2026 | 90 days |
| February 2, 2027 | 60 days |
| January 3, 2028 | 30 days |
Under the new approach, Project Zero plans to publish vulnerability details once the applicable deadline expires, even if the vendor has not released a patch.
Google has also removed most special extensions and exceptions from its standard disclosure process. Exceptions remain for certain situations, including vulnerabilities actively exploited in the wild, information already made public by others, and cases where a vendor requests disclosure.
The phased approach gives software companies time to improve their vulnerability management and patch-distribution processes before the deadline reaches 30 days.
Why AI Is Changing Vulnerability Disclosure
The main reason behind the policy changes is the rapid improvement of artificial intelligence in cybersecurity research.
Modern AI models can analyze large codebases, identify suspicious programming patterns, trace potential security weaknesses, and help researchers develop working demonstrations of vulnerabilities.
These capabilities can benefit defenders by making security research faster and less expensive. However, the same capabilities can also help attackers identify weaknesses and develop exploits.
Anthropic says the assumptions behind traditional disclosure policies are becoming less reliable. Earlier policies were designed around a world where software fixes moved relatively slowly and independent rediscovery of the same vulnerability was considered less likely.
That environment is changing.
If one researcher or AI model identifies a vulnerability, another model or attacker may independently discover the same flaw. The longer a vulnerability remains unpatched and undisclosed, the greater the opportunity for someone else to exploit it.
Google has raised a related concern: attackers can use AI to reverse-engineer software patches and understand the vulnerability that a fix addresses. This can reduce the practical protection offered by delaying public disclosure after a patch has been released.
Both companies believe that faster remediation is becoming increasingly important as AI-assisted vulnerability discovery improves.
What Happens When Developers Cannot Fix a Bug in 30 Days?
Shorter disclosure deadlines create additional pressure on software maintainers, particularly teams responsible for large open-source projects.
Some vulnerabilities are relatively straightforward to fix. Others require extensive testing, changes to underlying software architecture, coordination with multiple vendors, or updates to technical standards.
Anthropic acknowledges that certain vulnerabilities cannot realistically be resolved within 30 or 45 days. Examples include flaws affecting critical infrastructure software and vulnerabilities requiring changes to specifications or technical standards.
In these situations, the company says it will work with vendors and maintainers to establish a suitable remediation timeline.
Anthropic also says it will continue providing actionable information with vulnerability reports, including suggested patches or mitigations where possible. It plans to coordinate the volume of findings sent to individual projects so maintainers are not overwhelmed by large numbers of reports.
Google’s gradual transition similarly reflects feedback from vendors concerned that an immediate move to 30 days could create operational problems and potentially expose users to unfixed vulnerabilities.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.



