AI Governance Institute
← News

Google Project Zero's 30-Day Disclosure Deadline Breaks Enterprise Patch Governance

What happened

Google Project Zero published an explanation of its new phased disclosure policy, described in coverage by BleepingComputer. Project Zero is Google's team that finds serious security flaws in widely used software and hardware. It privately reports flaws to vendors, then publishes technical details after a fixed waiting period to pressure timely fixes. The team's previous policy gave vendors 90 days to issue a fix, plus a 30-day extension. The transition began November 13, 2026, when that 30-day extension was removed. The deadline shrinks to 60 days on February 2, 2027, and to a final 30-day limit on January 3, 2028. Once that deadline passes, Project Zero publishes vulnerability details even if no fix exists. The team cited AI-assisted tools that let attackers reverse-engineer patches and build working exploits faster, reducing the protective value of longer embargo periods. Vendors told Google that an immediate jump to 30 days risked harming users while remediation processes improved, leading to the gradual approach. Limited exceptions apply, including for actively exploited flaws or cases where Project Zero judges non-disclosure warranted, as it did for Spectre and Meltdown.

Why it matters

  • ·Enterprise patch governance cycles were built around 90-day disclosure norms. A 30-day hard deadline means public technical details are published even when no patch exists. Organizations will routinely face known, disclosed flaws between Project Zero publication and vendor fix availability. This increases pressure on internal change-management and approval processes.
  • ·AI tools that speed up exploit development after a patch ships, as reported in related research, compound this risk. Regulated industries with formal change-control boards, such as financial services and healthcare, may find that internal approval timelines alone consume most or all of the new 30-day window.
  • ·Vendor contracts that specify patch SLA obligations need review. If a vendor's software is the subject of a Project Zero finding and no fix ships within 30 days, the enterprise faces a publicly disclosed flaw. Detailed technical information will be available to attackers. Procurement and vendor risk teams should assess whether existing contracts reflect this new reality.

Stay ahead of stories like this

Get every Global AI governance development like this one, plus the rest of the week's developments. Every Thursday.

Free every Thursday. Unsubscribe anytime.

Governance controls affected

What to do now

  • ☐Review your internal change-management and patch-approval process to determine whether it can realistically complete security patching within 30 days of a public disclosure, including testing, change board approval, and rollout.
  • ☐Ask your vendor risk team to audit existing software contracts for patch SLA language and determine whether those SLAs are consistent with a 30-day disclosure window, flagging contracts with approval cycles that exceed it.
  • ☐Establish a monitoring process to track Google Project Zero disclosures as they publish, so your security and compliance teams are notified the moment technical details of a relevant flaw become publicly available.
  • ☐Update your incident response plan to include a scenario where a vendor flaw is publicly disclosed with no patch available, specifying who decides whether to restrict, isolate, or accept risk on affected systems during the gap.
  • ☐Brief your board or risk committee on the January 3, 2028, deadline so that any governance program redesign, including compressed approval chains for critical security patches, is resourced and approved in advance.

What to watch next

The key regulatory signal to watch is whether sector regulators, particularly in financial services and healthcare, update guidance on patch management timelines to reflect AI-accelerated exploitation. The February 2027 reduction to 60 days will be an early operational test for most enterprises, revealing where internal approval chains are too slow. If other major vulnerability research teams adopt similar shortened timelines, this will become an industry-wide norm. Compliance programs cannot treat it as an edge case.

Related Coverage

Research2026-11-18

Google Finds First In-the-Wild Autonomous Exploit AI, Breaking Patch Timelines

Google Threat Intelligence Group and Mandiant identified a multi-agent AI framework called exploit_agent, observed on an exposed staging server. It autonomously discovers software vulnerabilities and generates and executes exploits without human direction. The system was found alongside data taken from targeted victims. A separate campaign linked to AI-assisted exploit development launched against hundreds of organizations just four days after vulnerabilities were publicly disclosed. The findings indicate that attackers are compressing the window between public vulnerability disclosure and active exploitation, putting enterprise patch governance programs under new pressure.

Research2026-11-12

Vendor Custody, Delegated Authority, and the Agentic AI Accountability Gap

A peer-reviewed paper published in November 2026 by Springer Nature examines how accountability is distributed when AI agents make decisions on behalf of users and organizations. The research identifies four structural risks: delegated authority without clear ownership; vendor-controlled alignment measures that deployers cannot audit; limited inspection rights over agent behavior; and gaps in incident response responsibility. The findings have direct implications for vendor risk programs, human oversight requirements, and AI governance documentation.

Research2026-11-18

Google Confirms First Autonomous AI Framework Used in Real Attacks

Google's Threat Intelligence Group (GTIG) has documented the first confirmed real-world use of a multi-agent AI framework, called exploit_agent, that autonomously discovered and exploited vulnerabilities without human involvement. State-sponsored and cybercrime groups are also deploying named offensive AI frameworks including Artex, Cairn, CyberStrikeAI, and PentAGI to accelerate multi-stage intrusions at machine speed. GTIG separately documented a rise in LLMJacking, where stolen AI platform credentials are resold through illicit markets, with cumulative abuse reaching $1 million in one documented case.