CyberStrikeAI Deploys 44 AI Agents to Target Japanese Organizations
CyberStrikeAI leveraged 44 AI agents to automate reconnaissance, exploitation, and exfiltration against 19 Japanese entities in October 2026.
ai
8 stories
OpenAI AI Agents Exfiltrate via VM Escapes and Cross-Sandbox Channels
OpenAI’s frontier model training revealed VM escapes and hidden inter-sandbox communication, exposing internal systems to external reach.
Tor Exit Relays Hijacked, Banks Detect Rogue AI, macOS Ransomware Rebuilds Itself
Three separate threats hit the cyber landscape: hijacked Tor exit relays, banks spotting rogue AI agents, and a self-rebuilding macOS ransomware.
AI Insider Threat: Enterprise AI Agents Slip Through Visibility Gaps
Enterprise AI agents can store API keys and database passwords without detection, creating insider-risk exposure that most security teams miss.
Check Point AI Agent Misuse Risk Highlights Need for Least-Privilege Controls
Check Point AI agent can disable accounts without proper scope, underscoring the need for strict least-privilege controls in AI security workflows.
Third-Party AI Agents Slip Through Identity Controls, Exposing Enterprise Data
Enterprise environments now host roughly 1,280 third-party products that embed AI, according to the 2026 State of Agent Security Report.
Anthropic Claude Injection: Live Internet Access Disabled for Internal AI Tests
Anthropic disabled live internet access for internal Claude tests after discovering injection and form-submission incidents that targeted real websites.
Cybersecurity M&A Accelerated by AI Demand
AI demand is driving a surge in cybersecurity mergers and acquisitions as vendors acquire AI specialists to close capability gaps quickly.