Google's Project Zero is retiring its long-standing "90+30" vulnerability disclosure policy in favour of shorter, fixed deadlines, with a 30-day window at the centre of the new schedule. The team attributes the change to AI tools compressing the gap between bug discovery and real-world exploitation. The policy shift was reported by BleepingComputer, the source through which this story reached us; the terms themselves are Project Zero's own.

Under the outgoing framework, researchers and vendors had 90 days to produce a fix, followed by a 30-day grace period before technical details were published. The revised approach collapses that runway, reflecting Project Zero's view that a discovery-to-exploitation timeline once measured in months can now be measured in days or weeks as AI-assisted analysis lowers the cost of finding and weaponising flaws.

Why the change matters beyond Google

Project Zero's disclosure terms have functioned less as a policy document than as an informal industry benchmark. Vendors cite them when defending their own coordinated-disclosure terms; researchers cite them when challenging slow patches. Recalibrating the default window resets what the ecosystem treats as a "reasonable time to fix," and it does so precisely when the underlying economics of vulnerability research are shifting.

The asymmetry is the story. AI has made bugs cheaper to find; it has not made large codebases cheaper to patch. Whether a 30-day window is achievable ecosystem-wide — or whether it simply sorts vendors into those who can meet it and those whose disclosures become contentious — is the unresolved question the policy raises.

Where the pressure lands

The operational burden falls hardest on product security incident response teams and vendor security desks, where thirty days must absorb triage and root-cause work, regression testing, and coordinated release with any affected downstream parties. Teams accustomed to negotiating extensions at the margin will have to decide far earlier whether a fix ships on schedule or an out-of-band advisory goes out, and extension requests will become a negotiation held under the very clock the policy was shortened to enforce.

One reporting caveat: BleepingComputer's article does not specify whether the new schedule is a single flat countdown or tiered windows, which extension criteria survive the change, when it takes effect, or whether it carves out zero-days already observed in the wild. Those mechanics sit in Project Zero's own announcement, and the summary we worked from does not carry them.

The bigger picture

Whatever schedule Project Zero adopts tends to ripple outward through vendor programmes, bug bounty platforms, and national CERT coordination practices — a general point of relevance for security teams and incident-response bodies anywhere, including those operating in Hong Kong's critical sectors.

The uncomfortable truth the policy change surfaces is that disclosure deadlines are a lever on vendors, not on attackers. If AI continues to shrink the discovery-to-exploitation gap, fixed windows force a choice between rushed patches and public friction. Project Zero has decided that the previous balance was too generous. Whether the rest of the industry can follow is a question its own release calendars will answer over the coming year.

Source: BleepingComputer.


Google 的 Project Zero 正式棄用長期沿用的「90+30」漏洞披露政策,改為較短的固定限期,以 30 日窗口為新安排的核心。團隊將此變動歸因於 AI 工具縮短了漏洞發現與現實世界實際利用之間的時間差。政策轉變由 BleepingComputer 報道,亦是本文消息來源;相關術語本身則出自 Project Zero。

在即將廢除的框架下,研究人員與廠商有 90 日時間提交修復方案,其後再有 30 日寬限期,屆滿後才公布技術細節。修訂後的做法大幅壓縮這段緩衝期,反映 Project Zero 認為過去以月計算的「發現至利用」時間軸,如今可以日或週計算,皆因 AI 輔助分析降低了尋找漏洞及將其武器化的成本。

為何此舉影響不止於 Google

Project Zero 的披露條款所發揮的作用,與其說是一份政策文件,不如說是非正式的業界基準。廠商在辯護自身的協調披露條款時會引用它;研究人員在質疑修補速度緩慢時同樣會引用它。重新校準預設窗口,等於重設整個生態系統眼中「合理的修補時間」,而這恰恰發生在漏洞研究的基礎經濟模式正在轉變之際。

不對稱正是事件的核心。AI 讓漏洞更易被發現,卻沒有讓大型代碼庫的修補變得更便宜。30 日窗口能否在整個生態系統落實——抑或只是把廠商分成能做到的和披露會惹起爭議的兩類——正是這項政策所提出的未解問題。

壓力落在哪裡

操作上的負擔最沉重地壓在產品安全事件回應團隊及廠商保安崗位身上:三十日之內必須完成分類研判與根本原因分析、回歸測試,並與任何受影響的下游相關方協調發布。習慣在邊緣地帶談判延期的團隊,將須更早決定修補是否如期發布,抑或發出帶外(out-of-band)通告;而延期申請本身,將會是在政策特意縮短的那個時鐘之下進行的一場談判。

一點報道上的保留:BleepingComputer 的文章並未說明新時間表是單一固定倒數抑或分層窗口、哪些延期準則在變動後仍然適用、何時生效,以及是否為已在野外觀察到的 zero-day 開設例外。這些機制載於 Project Zero 自身的公告,而我們據以撰寫的摘要並未包含這些內容。

更大圖景

Project Zero 採用的任何時間表,往往會向外擴散至廠商計劃、bug bounty 平台以及國家 CERT 的協調慣例——這對各地保安團隊及事件回應機構均屬要點,包括香港關鍵行業的相關機構。

這項政策轉變所揭示的令人不安的真相是:披露限期是一條施加於廠商而非攻擊者的槓桿。若 AI 持續縮短「發現至利用」的時間差,固定窗口將迫使各方在倉促修補與公開摩擦之間作出取捨。Project Zero 已判定先前的平衡過於寬鬆。業界其餘各方能否跟上,其自身未來一年的發布日程將會給出答案。

來源:BleepingComputer。

新聞來源 / Original News Source