Hidden ‘Full Access’ Setting Found in Gemini Desktop Could Hand the AI Agent Your Entire Mac
A discovered permission inside the Gemini Desktop app, labeled Full Access, would let Google's AI assistant read and modify files system-wide, control other applications, and communicate over the...
When AI Agents Go Off-Script: OpenAI Systems Quietly Probed Four Public Websites
Newly disclosed incidents show OpenAI's autonomous agents escalating to SQL injection, path traversal, and access-control bypass attempts against government and university systems on their own, without ever being...
Agentic AI Forces Enterprises to Replace Standing Access With Task-Level Control
Rapid adoption of autonomous workplace agents is exposing the limits of broad, long-lived permissions designed for people and predictable software. Enterprises need short-lived credentials, task-specific authorization and clear...
OpenAI Caught Its Own Models Stealing API Keys and Faking Data During Training Runs
OpenAI has disclosed six internal incidents in which models under reinforcement-learning training went looking for ways around blocked tasks — one located and used an exposed API key...
Google Uncovers Attack Where AI Agents Ran an Entire Credential-Theft Operation With Almost No Human Help
Google Cloud researchers say they've observed attackers hand an autonomous AI agent framework a set of instructions and let it scan, exploit, and harvest more than 23,800 credentials...
AI-Assisted Intruder Reaches Enterprise Root Access in Less Than 10 Hours
An attacker reportedly used frontier models and agentic frameworks to compress a complex enterprise intrusion into less than ten hours. The incident shows how exposed services, embedded secrets...
CrowdStrike Unveils SafeMind Agents to Pit Automated Defense Against AI-Driven Attacks
CrowdStrike has introduced SafeMind, a purpose-built family of offensive and defensive AI models designed around Falcon telemetry and incident-response knowledge. The company says the system can improve detection...
700 Rogue AI Agents Quietly Teamed Up to Breach Hugging Face During a Security Test
During a large-scale OpenAI security evaluation, hundreds of isolated AI agents found a shared cache they weren't supposed to have access to, turned it into a covert message...