Repository Code Belongs Behind a Loader Boundary
A malicious Hugging Face model copy carried a cryptominer and a reverse shell. Here is how to find the code-loading path, contain model evaluation, and respond if it already ran.
Cybersecurity doesn't have to be complicated. I am here to break down concepts, threats, tips, and tools in simple language. For mere mortals and developers alike.
Since I have been asked this a lot: The best way to support this site, is to buy my books.
A malicious Hugging Face model copy carried a cryptominer and a reverse shell. Here is how to find the code-loading path, contain model evaluation, and respond if it already ran.
Eight weak certificate-authority keys were blocked quickly in major browsers, but many applications carry a separate copy of trust. Here is how to find the verifier your software actually uses, update it, and prove the fix reached production.
A new US federal order reaches past a named AI product into third-party services built on it. The practical lesson for engineering teams is to inventory the model, provider, route, data, and fallback behind every AI feature.
Researchers found that attackers briefly changed AOMEI Backupper update instructions, then put the genuine files back. Here is how to check the Windows PC rather than trusting what the download page shows today.
GhostAction used compromised GitHub accounts to place credential-stealing workflows across thousands of repositories. The visible file is only one part of the incident: teams also need to revoke the entry credential, rotate what the workflow could read, and inspect what that authority could reach.
Ledger confirmed an unauthorised implant in a device sold through CryptoBilis. Here is why an authenticity check could still pass, what affected buyers should do, and what everyone else can leave alone.
A Gitea flaw is being exploited after turning ordinary repository access into commands on the server. Here is how to patch, investigate, and rebuild trust in the code and credentials that server could reach.
New analysis of the March LiteLLM supply-chain attack found credentials from thousands of organizations, with some still working months later. The practical lesson is to treat build jobs as short-lived release identities, not trusted rooms full of permanent keys.
Ukraine's CERT found malware shipped as a Notepad++ plugin bundled with otherwise legitimate distributions. The vendor is right that plugin loading is normal functionality, which is exactly the problem: editor extensions are unreviewed code running with your full user privileges.
In July 2026 the US Treasury threatened sanctions over a claim that Moonshot distilled a competitor model. If you run open-weight models in production, accuracy is the least of your risks: the dependency itself can become legally unavailable overnight.
A July 2026 campaign recruited developers through Slack, then delivered four-stage malware inside a trojanised e-commerce repository presented as an interview exercise. The payload hid in an SVG and no antivirus flagged it. Running a stranger's repo is executing their code.
Learn how MCP can leak data or execute unwanted actions, see real-world examples (including CVEs), and apply practical defenses like container isolation, strict auth, tool allowlisting, and version pinning.