For five years, a vendor that got a bug report from Google's Project Zero could count on one thing: if it shipped a fix inside the deadline, the technical write-up would wait a while longer. That cushion is gone. In a post titled "Policy and Disclosure: 2026 Edition," dated Nov. 13, the research team said it is replacing its "90 + 30" disclosure policy with a stricter model it calls "Full X." Trade outlets covered the change on Nov. 16.

"Starting today, our policy is 'Full 90,'" the team wrote. Under the new rule, Project Zero publishes technical details of a vulnerability 90 days after it reports the bug to the vendor, whether or not a fix exists. The window then gets shorter on a fixed schedule: Full 60 from Feb. 2, 2027, and Full 30 from Jan. 3, 2028.

Some headlines have described this as a cut to 30 days. That is the end state for 2028. It does not apply today. For the next two and a half months the clock is 90 days. What changes right away is what happens when that clock runs out.

What exactly changed?

The old policy, in place since 2021, allowed vendors up to 90 days to fix a bug, plus an extra 30 days of "soak time" for patch rollout. Soak time is the gap between a patch being released and the bug's technical details being published. The idea was to give users time to install the update before attackers could read how the flaw worked. The old policy also had a grace period and a separate variant policy.

According to the Project Zero post and Infosecurity Magazine's summary, all three are gone: the 30-day soak time, the grace period and the variant policy. The new rule is short: the clock starts when the bug is reported, and details are published when it runs out.

Project Zero also says it will not use severity to set disclosure dates. A critical bug and a minor one get the same deadline.

Are there any exceptions?

Three, according to the Project Zero post, Help Net Security and Infosecurity Magazine:

  • Bugs exploited in the wild. These are handled under Google's application security vulnerability disclosure policy, not the Full X schedule.
  • Bugs whose details have already been published. If others have made the details public, Project Zero may publish its own at its discretion.
  • Vendor requests. Project Zero will publish details if the vendor it reported the bug to asks it to.

Separately, the post says Project Zero has occasionally granted exceptions on rare occasions, including for Spectre and Meltdown, and will continue to do so in the future. Help Net Security reads this as meaning hardware bugs may still get extra time.

Why is Google doing this now?

Project Zero's main argument is about AI. It says large language models (LLMs) have increased the speed at which attackers can find and exploit vulnerabilities, alongside the risk that multiple parties, including malicious ones, discover the same bug. "A vulnerability reported today presents a much higher immediate risk to users than one reported last year," Natalie Silvanovich wrote in the post. The post also says the benefits of delaying disclosure are declining because attackers are more likely to discover a bug during the embargo, though the team still sees some benefit in the short term.

The team also says AI-assisted discovery has raised the volume of reports: "We are finding several vulnerabilities a week in targets where we once found a few vulnerabilities a year." It adds that aspects of the old policy that required vendor negotiation, such as deciding whether a bug is fixed and whether a grace period applies, are "quickly becoming unmanageable."

The most important assumption sits behind the removal of soak time. Project Zero says LLMs have shortened the time attackers need to reverse engineer and exploit a bug from a patch, to the point that zero patch-to-exploitation time is "a reasonable, though imperfect, model of attacker behavior." In its words, soak time is no longer meaningful when attackers can almost immediately determine bug details from a patch, and it can put users at risk by creating a false sense of secrecy. The patch itself already shows attackers where the bug is.

The team still plans to track whether bugs are fixed and to say so when they are not, but fix status will no longer decide when details are published.

Why phase it in instead of going straight to 30?

Because vendors pushed back. Project Zero asked for feedback from the vendors it reports the most vulnerabilities to. Their main worry was that an immediate change could make unfixed bugs available to attackers while vendors were still improving their remediation processes. Vendors raised two other concerns. An aggressive policy could push them to fix Project Zero's reports ahead of more severe bugs reported elsewhere. And a deadline some vendors are not equipped to meet might lead them to accept that they cannot fix bugs in time and not try. The staged schedule of 90, then 60, then 30 days is meant to be, in the team's words, "challenging but realistic."

Ivan Milenkovic, VP of risk technology at Qualys, told Infosecurity Magazine that Project Zero had "removed the snooze button on responding to issues," and that for organizations, patching monthly now means being slower than disclosure.

Who wins and who loses?

Over the next 14 months, the largest effect will fall on vendors that tended to fix bugs right at the deadline. Under 90 + 30, a vendor that fixed a bug inside 90 days also got an extra 30 days of soak time before details were published. Under Full 90, day 90 is publication day with or without a patch. Under Full 30, a vendor that needs six weeks to build, test and ship a fix will see the technical details published before the fix exists.

For defenders, the effect cuts both ways. Public technical detail lets security teams write detections and judge exposure without waiting for a vague advisory. But it also means some bugs in software they run may be documented in public while no patch exists. Help Net Security's headline states the point plainly: bugs are disclosed after 90 days, fixed or not.

Project Zero's position is that secrecy is worth less than it used to be: if attackers can find the same bugs and turn patches into exploits almost at once, a long private window offers less protection to users than it once did. Whether the staged schedule pushes vendors to speed up will show in how they change their release cycles over the next year.

What It Means for Defenders

Who is affected: anyone running software from vendors that Project Zero researches. The policy change involves no single vulnerability, so there are no fixed versions to install from this news. The changes are to timing and process.

  • Expect public details on unpatched bugs. Starting now, a Project Zero bug can be published 90 days after reporting even when no fix has shipped. Your vulnerability management process should cover the case "technical details are public, vendor patch pending." That means compensating controls such as reducing exposure, restricting access to the affected feature or service, and closer monitoring, not only "patch when available."
  • Shrink your own patch window. With soak time gone, assume that once a Project Zero bug's fix is released, details may be public the same day or soon after. A 30-day patch cycle that worked when attackers needed weeks to reverse-engineer a fix is now a long exposure. Speed up deployment for internet-facing systems first.
  • Put the dates in your calendar. Full 60 starts Feb. 2, 2027. Full 30 starts Jan. 3, 2028. Plan staffing and change-window policy around those dates instead of reacting when the clocks shorten.
  • Ask your vendors. If a critical supplier usually fixes bugs near the 90-day mark, ask how it plans to meet a 60-day and then 30-day disclosure window, and what interim mitigation guidance it will give when a disclosure lands before a patch.
  • Don't sort by severity alone. Project Zero will not use severity to set disclosure dates, so a moderate bug may be published while a more severe one from another source stays quiet. When judging risk, consider whether technical details are public, not just the CVSS score.
  • Federal contractors and CMMC shops: review whether your documented flaw-remediation timelines still match how fast vulnerability details now become public, and record compensating controls for cases where details are out before a patch.

One unchanged rule is worth knowing: bugs already exploited in the wild are handled under Google's application security vulnerability disclosure policy, not Full X. Treat those as you always have, as active threats that need immediate attention.

Sources