Security Advisory

CVE-2026-58704: Google Pixel Cellular Modem Privilege Escalation (KEV)

Last updated:
  • CVE-2026-58704

CVE-2026-58704 is a permission-bypass logic error in the cellular modem firmware of Google Pixel devices that lets a radio-proximate attacker escalate privileges without any user interaction, and Google's own September 2026 Pixel bulletin says it may already be under limited, targeted exploitation.

Vulnerability summary

CVSS base score

8.8

High severity

EPSS score (FIRST)

0.6%

Probability of exploitation in the wild in the next 30 days

CISA KEV catalog

Yes

Listed as known exploited

Published

Vulnerability ID: CVE-2026-58704 · Source: NVD

Affected components

ComponentCPEVersion range
Google Pixel devices (cellular modem component)cpe:2.3:o:google:android:-:*:*:*:*:*:*:*All supported Pixel devices before the 2026-09-05 security patch level (fixed by updating to patch level 2026-09-05 or later)

What does this mean for affected devices?

CVE-2026-58704 is an improper authorization / incorrect privilege assignment (CWE-285, CWE-693) in the cellular modem firmware used by Google Pixel devices. NVD's description states: "In Cellular Modem, there is a possible permission bypass due to a logic error in the code. This could lead to remote (proximal/adjacent) escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation" — meaning an attacker within radio range of the device's cellular modem, not necessarily reachable over the open internet, can exploit the logic flaw to bypass a permission check and gain elevated privileges without any action from the user (NVD).

The CVSS v3.1 score NVD lists, 8.8 (high), vector CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H, is stored as a Secondary metric sourced to CISA's own Authorized Data Publisher (ADP) scoring — it is not Google's own CNA submission and not an NVD Primary score, though the adjacent-network attack vector matches Google's "remote (proximal/adjacent)" description of the bug (NVD).

Google's Android Security Bulletin for Pixel devices, dated September 2026, lists CVE-2026-58704 under the "Modem" component with internal reference A-484011314*, type "EoP" (Elevation of Privilege) and severity "High", and separately states: "There are indications that CVE-2026-58704 may be under limited, targeted exploitation." The asterisk on the bug ID indicates the fix is delivered through binary modem/baseband drivers rather than the AOSP source tree. All supported Pixel devices receive the fix at security patch level 2026-09-05 or later (Android Security Bulletin — Pixel, September 2026).

CISA added CVE-2026-58704 to the Known Exploited Vulnerabilities (KEV) catalog on 2026-09-16, with a required-action due date of 2026-09-19 (3 days), consistent with Google's own note about limited, targeted exploitation (CISA KEV).

As of 2026-10-27, FIRST scores the probability of exploitation at 0.0063 (49.750th percentile) — middling, unsurprising for a bug whose attack vector requires physical or radio proximity to the target device rather than being exploitable over the open internet (FIRST EPSS).

Because exploitation has already been observed in a limited, targeted fashion and the attack needs only radio proximity rather than a prior foothold on the device, every supported Pixel device should be updated to the 2026-09-05 security patch level or later as soon as it is available, with particular urgency for users who could be the target of focused surveillance.

How to fix this class of issueNo dedicated SoftWSP guide covers this weakness class; the fix is usually the vendor patch or firmware update referenced above. Our related guide explains how to plan and verify it: NIST SP 800-40 Rev. 4: enterprise patch management planning, explained.

Evidence

  • Permission bypass due to a logic error in the cellular modem (CWE-285/CWE-693) enables adjacent/proximal escalation of privilege with no additional privileges or user interaction required. — NVD
  • CVSS v3.1 base score is 8.8 (high), vector CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H, sourced to CISA's Authorized Data Publisher (ADP) scoring and stored by NVD as a Secondary metric (not Google's own CNA submission). — NVD
  • Android Security Bulletin (Pixel, September 2026) lists the bug under the Modem component (ref A-484011314*, type EoP, severity High), states it "may be under limited, targeted exploitation," and fixes it at security patch level 2026-09-05. — Android Security Bulletin — Pixel, September 2026
  • Added to the CISA Known Exploited Vulnerabilities catalog on 2026-09-16, required action due 2026-09-19. — CISA KEV
  • EPSS score is 0.0063 (49.750th percentile) as of 2026-10-27. — FIRST EPSS

Related advisories

Sources

  1. NVD
  2. Android Security Bulletin — Pixel, September 2026
  3. CISA KEV
  4. FIRST EPSS

Frequently asked questions

Does an attacker need internet access to exploit CVE-2026-58704?

No. The attack vector is adjacent/proximal, meaning an attacker needs radio-range access to the device's cellular modem rather than reachability over the open internet (per NVD).

Why does the bug ID in Google's bulletin have an asterisk?

The asterisk on bug reference A-484011314* indicates the fix is delivered through binary modem/baseband drivers rather than through the open AOSP source tree (per Android Security Bulletin — Pixel, September 2026).

Is the 8.8 CVSS score Google's own submission?

No. NVD stores it as a Secondary metric sourced to CISA's own Authorized Data Publisher (ADP) scoring, not Google's CNA submission, though the adjacent-network attack vector matches Google's own description of the bug (per NVD).

Has this vulnerability actually been exploited?

Google's own bulletin states there are indications CVE-2026-58704 may be under limited, targeted exploitation, and CISA's KEV listing — added with just a 3-day remediation window — is consistent with that note (per Android Security Bulletin — Pixel, September 2026; per CISA KEV).

Why is the EPSS score only 0.0063 despite the KEV listing?

This page attributes the middling 49.750th-percentile score to the attack vector requiring physical or radio proximity to the target device rather than being exploitable over the open internet, unlike many higher-EPSS KEV entries on this site (per FIRST EPSS).