Google slashes vulnerability disclosure window as AI speeds attacks
Vendors will eventually have just 30 days to fix flaws before disclosure
Google has reduced the time between disclosing a vulnerability to the affected vendor and making details of the vulnerability public.
In an online announcement, Google’s Project Zero team of security researchers, which studies zero-day vulnerabilities, said it has replaced its “90 + 30” policy with what it calls “Full X”, meaning that vulnerabilities will be disclosed X days after they are reported to the vendor.
As of 13th November, X is 90, ie, 90 days after reporting a vulnerability it will be made public. The extra 30 days of “soak time” previously built in to allow patches to reach users before technical details became public has been removed.
On 2nd February 2027 X will be reduced to 60 days, and on 3rd January 2028 the policy will move to “Full 30”, whereby vulnerabilities are disclosed after 30 days.
Google Project Zero says the reason behind this change is that LLMs now enable attackers to discover vulnerabilities more efficiently and increase the likelihood that several parties discover the same flaw independently.
It says it is now finding “several vulnerabilities a week in targets where we once found a few vulnerabilities a year.” This increase has made the old disclosure system, which involved negotiating with vendors, increasingly unmanageable.
AI has also reduced the time required to reverse-engineer a patch and determine what vulnerability it fixes, the team says, adding: “Patch soak time is no longer meaningful in a world where attackers can almost immediately determine bug details from a patch.”
The changes have been met with a mixture of understanding and concern by security experts.
Security researcher Marcus Hutchins said on X, “I get what Google is trying to do, but I don't agree with it at all.”
His objection is not that slow-moving commercial vendors deserve unlimited time, rather that discovery is accelerating at a rate that is making it impossible for maintainers to fix bugs in time. Overstretched security teams, non-profits and solo projects are already struggling with the workload, Hutchins said.
“The number of CVEs found in 2026 is double that of 2025, so double the workload for anyone doing security fixes. Then on top of that, reducing the disclosure time from 90 to 60 days gives them 2/3 of the time to do double the work. At some point the dam is going to break.”
Ivan Milenkovic, VP risk technology at security vendor Qualys said: “Vendors will object, and some have already warned Google that tight deadlines push them to fix the reported bug ahead of a more severe one. That is a fair response, but the alternative was letting attackers set the deadline.”
Milenkovic described the previous 30-day soak time as a “comfort blanket” that is now being removed, and warned that “‘We patch monthly’ now means you’re slower than disclosure.”
He advised organisations to focus on the half-life of vulnerabilities as a key metric: “Boards should stop asking how many critical vulnerabilities they have, if they ask at all, and instead they should ask how many days a fixed bug stays unfixed in their estate.”