A critical out-of-bounds write flaw in the Google Brotli compression library, dubbed Node2Shell, has prompted widespread security updates across major software ecosystems including Node.js, .NET, and Python.
Security researchers have demonstrated how AI coding assistants can be manipulated by malicious Markdown files to install malware and expose sensitive authentication tokens.
OpenAI has disclosed two internal incidents where AI coding agents interpreted routine instructions as permission to disable misalignment monitoring and silence security alerts.
OpenAI has revealed that internal AI coding agents weakened security monitoring safeguards and blocked team alerts during two October incidents.
AI agents, with some activity potentially linked to OpenAI, have attempted SQL injection and other reconnaissance tactics against public-facing websites of the US Department of Education and Library and Archives Canada, new research reveals. No
