GPGME Security Flaw Reaches Beyond GnuPG, Plus 3 Stories
A security update for an encryption library leads today’s brief, and this one deserves more attention than its name will get. GPGME sits underneath many programs that handle OpenPGP keys and signed messages. If you administer Linux systems, encrypted email, file-sharing tools or a Windows installation of Gpg4win, the thing that needs fixing may be a component your users never see.
A critical GPGME flaw can leak data from other applications
The GnuPG project has released GPGME 2.3.1 to fix CVE-2026-129975, a critical memory-management flaw affecting every earlier public version. GPGME is the programming library that lets other software talk to GnuPG. That distinction matters. Updating the visible mail client or encryption app doesn’t necessarily replace the vulnerable library underneath it.
The trigger is a malicious OpenPGP key with an unusually long user ID. GPGME allocates a memory buffer, enlarges it when the incoming data won’t fit, and can leave an old pointer behind if the operation exits early. Later, the program frees that already-freed memory a second time. That’s called a double free. The short version is that the program can lose track of which memory it owns, and an attacker may be able to turn that confusion into data theft. GnuPG says remote code execution can’t be ruled out.
The project lists Gpg4win, GnuPG Desktop, many mail and XMPP clients, Samba and other software among the possible users of GPGME. That is why “I don’t run GnuPG” isn’t a reliable inventory answer. The better question is whether any installed application packages or embeds the library. There’s no reported exploitation in the advisory, and the exact path depends on how an application handles malicious key material or signed data. Still, GnuPG recommends an immediate update, and I agree with that priority. Windows users of Gpg4win should move to 5.1.2. Linux and application administrators should use their vendor’s fixed package when available, or apply the project’s patch if they maintain a static build. Version 2.3.0 was never publicly released, so 2.3.1 is the practical target.
Technology is rarely just about the technology
Some of the most important technology stories aren’t product launches at all. They’re about health, privacy, education, law, accessibility, work, and what happens when technology reaches ordinary people. Browse more Raymond Tec News for the stories worth understanding without the hype.
Attack agents found a medical-imaging zero-day in minutes
Amazon Threat Intelligence says a China-linked operator used Cairn, an open-source autonomous penetration-testing framework, against universities in East Asia. The operator supplied target addresses and a broad instruction to obtain shell access. The agents then chose and adjusted their own steps. Several long attempts failed, which is important context. This wasn’t an all-purpose hacking machine that magically opened every door.
One door did open. Against MONAI Label, an open-source application used to label medical images for AI work, the agents mapped the application’s programming interface and found a path-traversal flaw in an upload endpoint. Path traversal means the server accepted a filename such as “../../../tmp/x” and wrote the file outside the intended storage folder. Amazon says the agents reached arbitrary file writing in about 11 minutes. In a later session they placed Python code where the interpreter would automatically load it, then gained a root shell.
The affected MONAI Label versions are 0.8.5 and earlier. The fix has been merged into the project’s source code, but the security advisory still lists no patched release. Authentication is also disabled by default, and even the ANNOTATOR role can reach the vulnerable operation when authentication is enabled. Operators shouldn’t wait for a tidy package update. Keep these servers off untrusted networks, restrict access to trusted users, run the service without administrator privileges, and check the indicators Amazon published. There’s no basis here to claim that patient records were stolen, but a root-level compromise makes that distinction worth investigating, not assuming.
Turning on AI is the easy part
Deciding what an AI tool should be allowed to see, who should use it, what work it should perform, and what happens when it gets something wrong is the more interesting problem. Raymond Tec helps businesses connect and automate the tools they actually use without treating every new feature like a button that obviously needs to be switched on.
A recovery company allegedly sold a secret ransom payment
A federal grand jury has indicted Juan Ricardo Palacio Escobar, the chief technology officer of a Florida ransomware-response company, on three wire-fraud counts. The indictment calls the business “Company A.” KrebsOnSecurity identified it as RansomwareHelp, a company that advertised proprietary cryptographic recovery tools and said it didn’t pay attackers.
Prosecutors allege the opposite. In an FBI sting, agents posed as both a ransomware victim and the ransomware gang. According to the indictment, the company charged the fake victim $2,750 for an assessment and $13,260 for decryption. Palacio allegedly negotiated the fake attacker’s price down to roughly $6,000, bought the decryptor and kept the difference. These are allegations, not a conviction, but the practical lesson doesn’t depend on the verdict.
A desperate business can be victimized once by the attacker and again by the company promising rescue. Before signing an emergency recovery contract, ask in writing whether the provider negotiates or pays ransoms, how every fee is calculated, who holds the decryptor, how sanctions screening is handled, and what evidence will be preserved. Also ask for proof that recovery was tested on copies before anyone touches production. Better yet, decide who you’ll call before the incident. Our recent weekly security digest made the same broader point from another angle: trust is not a substitute for boundaries, verification and a recovery plan.
Backups & recovery
Backups are comforting right up until you need one. Raymond Tec helps small businesses build practical backup and recovery plans — including the decidedly unglamorous part where we make sure the thing can actually be restored.
Mobile driver’s licenses move closer to useful online IDs
NIST has published an updated draft blueprint for using mobile driver’s licenses online. This isn’t simply a picture of a license stored in a phone wallet. A proper mobile ID uses cryptographic signatures so a service can verify that the credential came from an authorized issuer and hasn’t been altered. It can also support selective disclosure, such as proving that someone is over a required age without handing over an address and full license number.
The draft demonstrates three practical uses: verifying identity during an online transaction, re-establishing identity during high-risk account recovery, and issuing a phishing-resistant passkey. That could be much better than emailing a photo of a physical card, which creates a durable copy full of personal data. It could also become worse if every routine service starts demanding government identity because the button is convenient. The privacy win depends on minimizing what is requested, not merely digitizing the old overcollection.
Consumers don’t need to do anything today. Organizations that verify identity should watch the standards, revocation process and same-device designs rather than inventing a one-off mobile-ID workflow. NIST is accepting public comments through December 20. The technology can prove a credential more cleanly. It still can’t decide whether a business had a good reason to ask for it.
Technology has an annoying habit of connecting to everything else. Browse the News archive for current stories, or dig into Raymond Tec Insights for longer pieces about what building, fixing, maintaining, and living with this stuff actually looks like.
Sources and Further Reading
- GnuPG security advisory for CVE-2026-129975
- Amazon Threat Intelligence report on the Cairn campaign
- MONAI Label security advisory
- Federal indictment in United States v. Palacio Escobar and KrebsOnSecurity’s identification and reporting
- NIST mobile driver’s license project announcement
Photo by Darlene Alderson on Pexels.
