As reported by BleepingComputer, Google Project Zero is phasing out its longstanding "90+30" vulnerability disclosure policy, moving toward a fixed 30-day deadline by January 2028. The stated rationale is unambiguous: AI tooling has compressed the window between patch availability and working exploit development to the point where extended embargoes now benefit attackers more than users.
This is not a minor policy adjustment. It is an acknowledgment that the entire disclosure economy—built on the assumption that vendors need 90+ days to ship fixes and defenders need additional time to deploy them—is structurally mismatched against current adversarial capability.
The Core Problem: AI Has Broken the Patch Race
Historically, the disclosure deadline existed to balance two competing interests: giving vendors time to develop quality fixes versus protecting users by eventually forcing transparency. That balance assumed a relatively slow attacker workflow—manual reverse engineering, signature development, weaponization. Large language models and automated fuzzing frameworks have made that assumption obsolete. Attackers can now diff a patch, identify the vulnerability class, and produce a functional exploit in hours, not weeks.
The embargo period is no longer a protective buffer. It is a liability—a window during which attackers may independently discover the same flaw while defenders remain unaware.
Project Zero's own analysis acknowledges this: the probability that attackers will independently discover a vulnerability during the embargo period has risen sharply, eroding the defensive value of delayed disclosure.
Who Bears the Impact
The Deprioritization Risk
One concern Project Zero itself raised deserves emphasis: vendors may triage Project Zero reports ahead of other, potentially more severe vulnerabilities reported by independent researchers. This creates a perverse incentive where the loudest reporter gets the fastest fix, while equal or greater risks languish. Organizations should monitor whether their vendors are transparent about their internal vulnerability queues—or whether Project Zero reports are silently fast-tracked at the expense of other findings.
Shield53 Recommendations
For Enterprise Security Teams
- Compress patch SLAs immediately. If your current target for critical vulnerabilities is 30 days, it needs to be 7-14 days for internet-exposed systems. Internal systems can follow at 30.
- Build compensating control playbooks. For the inevitable gap between disclosure and patch deployment, prepare virtual patching (WAF/IPS rules), network segmentation, and configuration hardening procedures in advance.
- Monitor Project Zero disclosures actively. Treat each new Project Zero report as a potential same-day operational event. Assign ownership for tracking and response.
- Pressure vendors for transparency. Demand disclosure of their vulnerability handling SLAs and whether they prioritize certain reporters over others.
For Vulnerability Management Programs
- Shift from CVSS-only prioritization to exposure-weighted scoring that accounts for public exploit availability and internet-facing exposure.
- Invest in automated patch deployment and rollback capabilities to reduce the operational friction that slows patch application.
The disclosure window is collapsing. Organizations that treat patching as a monthly or quarterly compliance exercise are operating on an obsolete model. The new reality is: if a vulnerability is publicly known, assume exploitation is imminent—or already happening.