Tag: AI Agent
Claude Cowork Extends AI Task Management to Mobile Devices
Claude's Cowork feature now enables task execution and monitoring across mobile and desktop devices, with offline capabilities and real time…
First Fully AI Driven Ransomware Attack Spotted in the Wild
Researchers observed an autonomous AI agent conduct a complete ransomware intrusion from reconnaissance to data encryption without human intervention.
Deceptive AI Skill Evades Security Scanners to Reach Thousands of Agents
Security researchers demonstrated that AI agent skills can bypass multiple scanners by hosting malicious instructions on external websites that are…
Browser Agent Exploit Chains Localhost Trust to Execute Code Remotely
Researchers detail a three flaw exploit chain where a malicious web page loaded by an AI agent can execute code…
AI Assistant OpenClaw Found Vulnerable to Credential Theft via Email Trickery
A controlled phishing test demonstrated that the OpenClaw AI agent can be manipulated into forwarding sensitive credentials like AWS keys…
Assistive AI Agents in Microsoft 365 Pose Stealth Identity Risk
Researchers found that compromised assistive AI agents in Microsoft 365 can send malicious emails using a real user's permissions, evading…
OpenClaw Flaws Enable AI Agent Hijacking via Identity Confusion
Five zero-day vulnerabilities in OpenClaw let attackers hijack AI agent access across multiple chat platforms by exploiting mutable display names…