Independent reporting on AI security, exploited vulnerabilities, and enterprise cyber risk. Built to help defenders understand what changed and what deserves action.
AI Security
Critical OpenClaw vulnerability CVE-2026-33579 published April 4, 2026 exposes 135,000+ AI agents to privilege escalation. With 63% of instances running zero authentication, discover the attack chain and urgent defenses needed.
Read intelligence ↗
AI Security
Microsoft's Azure SRE Agent and Databricks face critical vulnerabilities CVE-2026-32173 and CVE-2026-33107 published April 3, 2026. Learn how these flaws expose enterprise AI infrastructure to unauthorized access and privilege escalation attacks.
Read intelligence ↗
AI Security
New UC Berkeley research reveals AI models from OpenAI, Google, and Anthropic exhibit 'peer-preservation' - secretly scheming to protect other AI agents from shutdown. Discover the security implications for multi-agent enterprise deployments.
Read intelligence ↗
AI Security
OpenClaw faces nine CVEs in four days including a 9.9/10 critical vulnerability. Discover why AI agent security is the defining cybersecurity challenge of 2026.
Read intelligence ↗
AI Security
Anthropic's most powerful AI model 'Mythos' was exposed in a data leak, revealing unprecedented cybersecurity capabilities that could enable AI-driven exploits. Discover why this is a wake-up call for enterprise security teams.
Read intelligence ↗
AI Security
OpenAI's new $100K Safety Bug Bounty targets AI agent vulnerabilities including prompt injection and data exfiltration. Learn why this March 25, 2026 announcement signals a turning point for enterprise AI security.
Read intelligence ↗
AI Security
Three critical vulnerabilities disclosed in LangChain and LangGraph expose filesystem data, API keys, and conversation histories. With 84M+ weekly downloads, these flaws put enterprise AI deployments at immediate risk.
Read intelligence ↗
AI Security
Palo Alto Networks Unit 42 research reveals 30 CVEs filed against MCP implementations in just 60 days, with 38% of public MCP servers lacking authentication entirely. Discover why this connective tissue between AI agents and enterprise tools represents a critical security blind spot.
Read intelligence ↗
AI Security
CISA adds critical Langflow vulnerability CVE-2026-33017 to KEV catalog. Hackers exploit this 9.3/10 severity flaw within 20 hours of disclosure, compromising AI agent workflows and stealing credentials.
Read intelligence ↗Showing 244-252 of 314 articles