Google Chrome EUVD-2026-93760

Exploitation requires that the attacker has ALREADY compromised a Chrome renderer process via a separate vulnerability - this is the primary gating prerequisite and the reason CVSS uses PR:L rather than PR:N. Given that foothold, the attacker must deliver a crafted PDF file that triggers the missing authorization check (CWE-862) in the Workers implementation to cross the site-isolation boundary. The vulnerability is fixed in Chrome 155.0.8059.39, so only earlier versions are affected; no end-user interaction beyond the PDF handling is described. The key limiting factor is that this is a chained sandbox/site-isolation bypass, not a standalone remote exploit, which substantially reduces real-world exploitability despite the network attack vector.

Signals here are consistently moderate rather than critical, and they align well with each other. The CVSS 3.1 base score is 8.1 (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:N), but the PR:L and the description's explicit precondition - 'a remote attacker who had compromised the renderer process' - show this is not a standalone remote vulnerability. It is a site-isolation/sandbox-boundary bypass that only becomes useful as the second link in an exploit chain, after a separate renderer-compromise bug is already in play. The vendor itself rated the Chromium security severity as Medium, which contrasts with the High CVSS score; the gap is explained by the required pre-compromise. Corroborating lower-tier signals: EPSS is 0.17% (5th percentile), it is NOT on the CISA KEV list, SSVC reports Exploitation: none and Automatable: no, and no POC is referenced. The one 'high' signal is SSVC Technical Impact: total, which is consistent with C:H/I:H (full cross-site data access once the boundary is bypassed). Net assessment: this is a real defense-in-depth fix worth deploying promptly given Chrome's reach, but it is not an independently exploitable, drive-by priority - it requires a chained renderer compromise and shows no evidence of active or automatable exploitation.

An attacker who has already exploited a separate bug to gain code execution inside a Chrome renderer process serves or references a specially crafted PDF file, abusing missing authorization in the Workers implementation to escape the per-site boundary. This lets the compromised renderer read and tamper with data belonging to other sites/origins that site isolation is supposed to keep separate. No public exploit is identified at time of analysis, and the attack is only meaningful as a chained stage after an initial renderer compromise.