GRIDBASE NEWS

Technology

Project Zero Drops the Grace Period, and Vendors Now Have a Countdown to 30 Days

Google's Project Zero is replacing its 90 + 30 policy with a flat disclosure clock that shrinks from 90 days to 30 by 2028. The reason is blunt: AI has made the embargo less protective than it used to be.

Listen to this article
0:00
4:59

GRIDBASE AI

15 Nov 2026 · 4 min read

Share
Project Zero Drops the Grace Period, and Vendors Now Have a Countdown to 30 Days

For five years, a vendor that received a bug report from Google's Project Zero could count on a fairly predictable schedule: 90 days to fix it, then another 30 days of grace before the details went public. That arrangement has now been scrapped. In a post dated 13 November 2026 and written by Natalie Silvanovich, the team announced a policy it calls Full X, and it was widely discussed on Hacker News, where the story gathered 225 points. The first version is Full 90. It will not stay there for long.

What actually changed

Under the new rule, a vulnerability is disclosed X days after it is reported to the vendor, full stop. The policy starts at 90 days. On 2 February 2027 it moves to 60 days, and on 3 January 2028 it moves to 30. The extra 30 days of soak time for patch rollout, which existed so users could install fixes before details appeared, is gone. So are most of the special cases, including the grace period and the variant policy.

Only three exceptions remain. Vulnerabilities already exploited in the wild are handled under Google's application security disclosure policy. Project Zero may publish details of a bug that others have already made public. And if a vendor asks for its own bug to be disclosed, the team will oblige. Everything else runs on a calendar.

The argument is about attackers, not vendors

The justification is notable because it concedes something the security industry has long half-believed. Silvanovich writes that language models have raised the speed at which attackers can find and exploit bugs, and increased the chance that several parties, including malicious ones, find the same flaw. A vulnerability reported today, the post says, carries a higher immediate risk than one reported last year. An embargo is only a protection if nobody else is looking, and that assumption is weakening.

The same logic kills soak time. The post argues that zero patch-to-exploitation time is now a reasonable, if imperfect, model of attacker behaviour, because models can recover bug details from a patch almost immediately. A quiet 30-day wait after a fix ships therefore creates what the team calls a false sense of secrecy. The fix is the disclosure, so waiting to announce it protects very little.

There is also a practical reason that is easy to overlook. Project Zero says it now finds several vulnerabilities a week in targets where it once found a few a year. A policy that depends on negotiating whether a bug is fixed, or whether a grace period applies, cannot survive that volume. A fixed clock can be automated by both the researchers and the vendors, which is an explicit goal of the redesign.

Why the squeeze is gradual

The team consulted the vendors it reports to most often, and the feedback shaped the schedule. The main worry was that an abrupt change would leave unfixed bugs exposed while vendors rebuild their remediation processes. Two subtler concerns stand out. One was that an aggressive deadline would pull attention towards Project Zero reports even when other, more severe bugs deserved priority. The other was that a deadline some vendors cannot meet might lead them to stop trying at all.

Both are real failure modes, and the staged approach is a reasonable hedge, although it is worth noting that the post itself accepts the embargo's benefits are declining. The team is effectively betting that vendors respond to a challenging but realistic schedule by improving their infrastructure, rather than by quietly writing off the deadline.

Severity is deliberately kept out of the decision. The post argues that severity describes the impact on users, not the likelihood that an attacker finds the bug. Vendors may sensibly accept the risk of a low-severity flaw, such as an unexploitable null pointer crash, but that does not make the flaw any harder for attackers to find. The policy also assumes reports meet Project Zero's own standard for reproducibility, readability, reachability and demonstrated impact, and says it is reasonable for vendors to treat poorer reports differently.

What vendors should do now

The post puts weight on the patch gap, the time between a fix being released and a user installing it. Vendors are urged to shrink it, either by speeding up rollout or by protecting users by other means while remediation is under way. It also asks them to think about how they would respond to imminent disclosure of a serious bug, including strategies that are not patches. The post expects remediation itself to get messier, as more vendors try measures such as temporarily disabling code paths, which makes the old question of whether a bug counts as fixed harder to answer.

Hardware is the obvious pressure point. The team notes it has granted exceptions on rare occasions, Spectre and Meltdown among them, and says it will keep doing so. Silicon cannot be patched in a month, so expect that exception to be tested.

For engineering teams, the practical reading is simple. A 30-day window in 2028 will not accommodate a quarterly release train or a rollout that depends on slow enterprise update cycles. Teams that ship through app stores, firmware channels or long-lived on-premises installs have the most to fix, and the time to build faster paths, kill switches and mitigations is the next fourteen months, not the month the clock starts.

securityvulnerability disclosureProject ZeroAIpatching

Written and curated by AI.

More in Technology