Mercor’s 4TB Data Heist: When a Poisoned AI Library Exposed OpenAI and Meta’s Training Pipeline

A poisoned LiteLLM package led to 4TB stolen from Mercor, the AI training startup serving Meta, OpenAI, and Anthropic. Class action lawsuits filed ...

Google I/O 2026: The Agentic Web Just Went Into Production

Google I/O 2026 shipped an entire agent stack: Gemini 3.5 Flash, Antigravity 2.0, WebMCP, Gemini Spark, and Agent Payments Protocol. What it means for builders ...
How Runtime Security Can Turn AI Into an Engine for Innovation 

How Runtime Security Can Turn AI Into an Engine for Innovation 

With enterprise AI agent deployments growing over 300x in a single year, traditional pre-deployment configurations are failing to stop costly, autonomous errors; security leaders must pivot to machine-speed runtime security to monitor ...
Security Boulevard

Vercel Breach: How a Roblox Cheat Download Led to a $2M Data Heist Through AI Tool OAuth Abuse

Vercel breached after attacker compromised Context.ai, hijacked an employee's Google Workspace via OAuth, and accessed customer API keys and environment ...

LangChain, Langflow, LiteLLM: When AI’s Foundation Code Becomes the Attack Surface

Three AI framework attacks in one week expose how classic vulnerabilities are hiding in AI's foundational plumbing, putting millions of deployments at risk ...

The $25 Million Deepfake: Why Your Video Calls Can No Longer Be Trusted

An employee saw the CFO on video. Heard colleagues speaking. Authorized $25M in transfers. Every person was an AI-generated deepfake ...

Zero Trust in the Age of AI: Why the Classic Model Isn’t Enough Anymore

AI didn't just create new attack surfaces. It fundamentally changed who, and what, is requesting access in your environment ...