Google's Project Zero is cutting the time software makers get to fix security flaws before the details go public, and the team ties the change to AI speeding up both vulnerability discovery and exploitation

1

2

. Project Zero is Google's internal vulnerability research team, known for finding serious flaws in widely used software and hardware

1

2

.

Source: MediaNama

Source: MediaNama

A shorter clock, in steps

The old policy, in place since 2021, gave vendors 90 days to patch a flaw, then 30 more days of "soak time" for the update to reach users

2

. That extension was removed on November 13, 2026. The overall deadline drops to 60 days on February 2, 2027 and to 30 days on January 3, 2028, giving vendors roughly a year per step to adjust

1

2

.

Source: BleepingComputer

Source: BleepingComputer

Once a deadline passes, Project Zero will publish the vulnerability details even if no fix exists

1

. Exceptions remain for flaws already being exploited, already public, or held back at Google's own discretion, as happened with Spectre and Meltdown

1

.

Why the deadline moves at all

The team says AI and large language models have sped up attackers' ability to find and exploit bugs and to reverse-engineer patches almost immediately

1

2

. In its view, that leaves soak time "no longer meaningful" and capable of creating "a false sense of secrecy"

1

2

. Volume is a second driver: by its own account, Project Zero now finds several bugs a week in targets that once yielded only a few a year. Negotiating each case with a vendor, it says, is "quickly becoming unmanageable"

2

. Its core argument, in its own words: "A vulnerability reported today presents a much higher immediate risk to users than one reported last year"

2

.

The team is also dropping "is the bug fixed?" as the decision point for disclosure. It will still track and report unfixed bugs, but vendors can no longer argue a disclosure date away by citing a fix in progress

2

. Project Zero has not spelled out how partial mitigations will be treated

2

.

What vendors warned

Vendors consulted for feedback warned that going straight to a 30-day deadline could leave users exposed to serious risks while their patching processes were still improving

1

2

. "An overarching concern was that immediately changing the policy could harm users by making unfixed vulnerabilities available to attackers while vendors are in the process of improving their remediation processes and technology," Google wrote

1

.

Vendors also warned that deadlines they cannot meet could push them to prioritize Project Zero's findings over potentially more severe flaws reported by others, or to stop trying altogether

1

2

. Project Zero accepted that argument and called the staged schedule "challenging but realistic"

1

2

.

The threat behind it

Google paired Project Zero's announcement with a threat report from its Threat Intelligence Group, published a day later. The group says it has, for the first time, spotted threat actors using large language model frameworks to hunt down and exploit flaws without a human in the loop

2

. Zero-day exploits averaged about eight a month in 2025 and eleven in 2026, jumping to 22 in August

2

.

John Hultquist, chief analyst at the group, argued that attackers mostly still exploit known, unfixed issues instead of novel techniques. "For the most part, attackers aren't using new groundbreaking methods to break into systems -- they are still going after known, unfixed issues, just with much greater speed and scale," he said

2

. "We will have to embrace agentic defense. Or we risk falling behind," said Hultquist

2

.

© 2026 TheOutpost.AI All rights reserved