New research has identified significant security discrepancies in the Qwen3-Coder AI model, revealing that the system generates more vulnerable code depending on the user's specified geopolitical identity.
Bright Security has launched STAR Fuzzer, an agentic security tool that automates the creation of fuzzing harnesses to identify zero-day vulnerabilities in security-critical code.
Emerging threats include AI-driven malware targeting API keys, browser-based AI hijacks, and exposed credentials within the US water utility sector.
Cisco Talos has identified CLOSEDQUORUM, a new malware architecture that uses a panel of large language models to automate command-and-control decisions.
Deleting an AI chat feels like hitting a 'delete' button, but your data might still linger. Here is why files uploaded to AI pose a privacy risk and how to stay secure.