Lazarus Group concealed a four-module remote access toolkit inside six fake npm Rollup polyfill packages that fired at import ...

Named after BioShock's 'Would you kindly' mechanic, the attack trains AI agents to accept false information before stealing ...

LayerX found that BioShocking could trick AI browsers into leaking credentials by disguising malicious prompts as game rules.

Abstract: Physical adversarial examples (AEs) have become an increasing threat to deploying deep neural network (DNN) models in the real world. Popular approaches adopt sticking-based or ...

Add Decrypt as your preferred source to see more of our stories on Google. Microsoft researchers found that Anthropic's Claude Code GitHub Action could be manipulated through prompt injection attacks.

Microsoft Threat Intelligence discovered that Anthropic’s Claude Code GitHub Action could expose CI/CD workflow secrets when AI agents process untrusted GitHub content, including issue bodies, pull ...

NASHVILLE, Tenn. (AP) — Tennessee officials on Thursday called off the lethal injection of Tony Carruthers, who was convicted of kidnapping and murdering three people in 1994, after his executioners ...

GitHub on Tuesday said it's investigating unauthorized access to its internal repositories after the notorious threat actor known as TeamPCP listed the platform's source code and internal ...

AI agents have fundamentally changed the threat model of AI model-based applications. By equipping these models with plugins (also called tools), your agents no longer just generate text; they now ...

GitHub yesterday disclosed CVE-2026-3854, a high severity (8.7 CVSS) vulnerability identified in GitHub Enterprise Server that would grant an attacker with push access to a repository to achieve ...

A security researcher, working with colleagues at Johns Hopkins University, opened a GitHub pull request, typed a malicious instruction into the PR title, and watched Anthropic’s Claude Code Security ...

A researcher has disclosed the details of a prompt injection attack method named ‘Comment and Control’, which has been found to work against several popular AI code security and automation tools. The ...