Italy's ACN warned on 2 November about four flaws rated 8.5 under CVSS 4.0. Three in FTS5 and the session extension were fixed in 3.54.0; the .recover command injection is listed through 3.54.1 with no fixed release.
GPGME 2.3.1 fixes a critical double free reachable through a crafted OpenPGP key
EditorialBy TrustList Editorial
GnuPG released GPGME 2.3.1 on 20 November 2026 to fix CVE-2026-129975, a double free triggered by a key with a very long user ID; Gpg4win, GnuPG Desktop, most mail clients and Samba use the library.
- Germany
- Cybersecurity
- Vulnerability Management
- Patch Management
- +1 more
About GPGME 2.3.1 fixes a critical double free reachable through a crafted OpenPGP key
GPGME 2.3.1 fixes a critical double free reachable through a crafted OpenPGP key
20 November 2026: The GnuPG project has released GPGME 2.3.1 to fix what it calls a critical security bug found in all earlier versions, CVE-2026-129975. GPGME is the library that most applications use to talk to GnuPG, so the fix matters far beyond GnuPG itself: Gpg4win, GnuPG Desktop, GnuPG VS-Desktop, most mail clients, many XMPP clients, Samba and Flatpak are named as users. New builds of Gpg4win, GnuPG Desktop and GnuPG VS-Desktop were released the same day.
Not yet independently verified. The flaw, affected software, fix and timeline come from the GnuPG announcement by Werner Koch of 20 November 2026 (read on 22 November 2026). It gives no CVSS score. The stated impact is data exfiltration; remote code execution is described only as something that cannot be ruled out. We will update this when it can be confirmed, and remove this note.
How the bug works
An OpenPGP key with a crafted, very long user ID can trigger a double free. The fault sits in two internal GPGME functions that read status lines and key listings from gpg. They start with a 1024-byte buffer and reallocate it when more room is needed, but only store the new pointer back in the context at the end of the function. If the function returns early, for example on a read error, the context keeps the stale pointer and the memory is later freed a second time. Because other allocations happen in between, the usual double free checks in memory allocators may not catch it.
The announcement says the bug can be used to exfiltrate data from applications, for example by sending signed data, and that remote code execution "can't be ruled out". It advises an immediate update.
Disclosure brought forward by five days
The flaw was found by Johan Carlsson, and GitLab told the GnuPG project on 12 November 2026. A coordinated disclosure was set for 25 November. Because GitLab had to ship an earlier update for a separate bug, the GPGME disclosure was moved forward to 20 November.
What is affected and what to install
Every program using a GPGME version before 2.3.0 is affected, whether it loads GPGME as a shared library or DLL or has it statically linked. Version 2.3.0 was never released to the public. The fix is:
- GPGME 2.3.1 (dated 19 November 2026 in its release notes), which also adds constants for the post-quantum ML-KEM and ML-DSA algorithms;
- for Windows, the GPGME DLL from the updated Gpg4win;
- for systems that cannot update, a patch attached to the announcement that the project says should apply even to very old versions.
The C++ and Qt bindings, GPGMEPP and GPGMEQT, do not need an update.
Steps for IT and security teams
- Update Gpg4win, GnuPG Desktop and GnuPG VS-Desktop on managed Windows machines.
- Apply the GPGME updates from your Linux distribution as they arrive, and check servers that run Samba or mail tools that verify signatures automatically.
- Ask the vendors of mail clients and other tools that bundle or statically link GPGME when they will ship 2.3.1; static builds stay vulnerable until they are rebuilt.
- Until then, be wary of importing keys from unknown senders or key servers on unpatched systems.
Sources
- gnupg-announce: GnuPG Security Advisory for GPGME (T8557, CVE-2026-129975), 20 November 2026
- GnuPG task T8557, 20 November 2026
Categories & features
- Germany
- Cybersecurity
- Vulnerability Management
- Patch Management
- Data Security
TrustList Weekly
The week in software and IT, in one email
The news that matters to buyers, new rankings and our own research. Every Thursday, free, and easy to leave.
You may be interested in
Attackers are encrypting NAS devices through a flaw Synology fixed in December 2024 and points to advisory SA-24:20, whose worst bug, CVE-2024-10441, allows remote code execution without a login.
AWS bulletins of 29 September and 1 October 2026: Powertools for AWS Lambda (Python) 3.6.0 to 3.34.0 could return unmasked values on error (fixed in 3.35.0, no workaround), plus fixes for Amazon Ion Python, security-agent-mcp-server and GluonTS.
Kiteworks published 125 security advisories on 30 September 2026, 12 rated critical, for Core, the Email Protection Gateway, Secure Data Forms and MFT Server. The most severe, a gateway code-injection flaw, is fixed in 9.4.1; update to 9.5.1.
Two of the nine run code on a viewer that connects to a hostile machine or opens a crafted recording; six let a local user reach SYSTEM or root through the Monitoring and Patch Management add-ons.
A captured login response could be replayed to get into a Mendix app that accepts identity-provider-initiated sign-on. Update the SAML module to 4.2.4, or 3.6.29 on Mendix 9.
Four of the 16 new notes are critical, including code injection in S/4HANA and broken access control in NetWeaver Visual Composer; SAP reports no exploitation so far.
TeamViewer’s bulletin TV-2026-1010 of 29 September 2026 fixes five flaws in Full Client and Host, rated up to CVSS 8.8, one a session access-control bypass that could allow code execution. Update to 15.82 or a fixed maintenance release.
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- More GermanyThe ranking for this subject
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- AI ModelsBenchmark scores and community ratings for every major model.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.