CVE-2026-42208 exploited within 36 hours of disclosure, exposing LiteLLM credentials, risking cloud account compromise.

How indirect prompt injection attacks on AI work - and 6 ways to shut them down ...

Security leaders must adapt large language model controls such as input validation, output filtering and least-privilege ...

Google has analyzed AI indirect prompt injection attempts involving sites on the public web and noticed an increase in ...

People who have had a heart attack, stroke, or serious circulation problem in their legs, and who also carry excess weight, can now be offered a weekly injection to help protect them from a further ...

In today's security landscape, some of the most dangerous vulnerabilities aren't flagged by automated scanners at all. These ...

A prompt injection flaw in Google’s Antigravity IDE turns a file search tool into a remote code execution vector, bypassing ...

A new gene therapy is giving people born deaf the chance to hear, often within just weeks. In a small but groundbreaking study, researchers delivered a working copy of a key hearing gene directly into ...

Agentic AI tools present the possibility of substantial efficiency gains for legal teams, but the risks they pose require ...

AI's danger isn't that it's creating new bugs, it's that it's amplifying old ones. On March 10, 2026, Microsoft patched ...