Apple iOS EUVD-2026-104734

Exploitation requires an attacker to already be executing code locally on the device as a malicious app running inside the app sandbox (AV:L, PR:L) - there is no remote or network path. The specific trigger is a race condition (CWE-362): the attacker must repeatedly exercise a timing-sensitive operation to win a narrow window before Apple's added validation check completes, which is reflected in the high attack complexity (AC:H) and makes exploitation unreliable and non-deterministic. No user interaction beyond initially installing/running the malicious app is needed (UI:N). Limiting factors for defenders: the attacker must first get the app onto the device (constrained by App Store review, notarization, and sideloading restrictions), must win a non-deterministic race, and only pre-27 versions of iOS/iPadOS/macOS are affected - updating to version 27 fully remediates.

This is a genuine high-priority patch-soon issue, though not an emergency. The CVSS 3.1 vector (AV:L/AC:H/PR:L/UI:N/S:C/C:H/I:H/A:H, base 7.8) describes a locally-exploitable race condition (CWE-362) that requires an attacker to already be running code with low privileges on the device - i.e. a malicious app that the user has installed - and to win a timing window (AC:H). The payoff is significant: a scope change (S:C) with high confidentiality, integrity, and availability impact, reflecting a sandbox escape that grants access to resources beyond the app's own sandbox. On the exploitation-signal side there is no CISA KEV listing, no EPSS score provided, no SSVC decision data, and no public proof-of-concept in the supplied intelligence, so there is no public exploit identified at time of analysis - this is NOT confirmed active exploitation. The practical risk is therefore gated by two real barriers: the attacker must first get a malicious app onto the device (App Store review, notarization, or sideloading limits this), and must reliably win a race condition. The 'Information Disclosure' tag understates the issue - a sandbox escape with I:H/A:H is a broader integrity/availability compromise, not merely disclosure. Signals are consistent (local vector, requires attacker-controlled app, high impact); the main missing data are EPSS and any SSVC/automatable assessment.

A user installs a malicious or trojanized app that passed store review or was sideloaded. The app repeatedly triggers a timing-sensitive operation to win a race condition and break out of the app sandbox, then reads protected user data and tampers with system resources it should never reach. No public exploit code was identified in the supplied data, and the race window (AC:H) makes exploitation non-trivial but feasible with repeated attempts.