Apple macOS CVE-2026-86933

Exploitation requires an attacker-controlled or compromised application already executing locally on the macOS host (AV:L, PR:L) under a user account; there is no remote or unauthenticated path. The flaw lets such an app bypass sandbox restrictions to access the user's Contacts data, normally gated by Apple's TCC privacy controls. Limiting factors: the attacker must first achieve local code execution via an app the user runs; impact is confined to reading contact information (C:L, no integrity or availability impact); and the issue is remediated in macOS Golden Gate 27, so only unpatched systems running untrusted apps are at risk. No user interaction with the exploit itself is required once the app runs (UI:N).

This is a genuinely low-priority issue across nearly every signal. The CVSS 3.1 base score is 3.3 (Low) with vector AV:L/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:N - local attack vector, low complexity, but requiring already-present low privileges (a running app) and yielding only limited confidentiality impact (read access to contacts) with no integrity or availability impact. EPSS reinforces this with a 0.14% exploitation probability (3rd percentile), and CISA's SSVC decision framework classifies Exploitation as 'none', Automatable as 'no', and Technical Impact as only 'partial' - all consistent with a non-urgent, non-wormable local information-disclosure flaw. There is no CISA KEV listing and no public exploit identified, so there is no evidence of real-world exploitation. All signals (CVSS, EPSS, SSVC, KEV, POC) agree and do not conflict: this is a legitimately low-risk, defense-in-depth sandbox hardening fix rather than a high-severity priority. The only caveat is that the vulnerability requires a malicious or compromised app to already be running locally, which is a meaningful precondition that further bounds real-world risk.

A user installs or runs a malicious or compromised sandboxed application on macOS; the app exploits insufficient sandbox restrictions to read the user's contacts data (names, phone numbers, emails) without the user's consent or a TCC prompt. No public exploit code has been identified, and exploitation requires local code execution via the app, so this is an information-disclosure escalation rather than remote compromise.