Deadbugz: The MCP Attack That Waits 3 Calls Before Striking
Deadbugz filed 23 GitHub PRs with a malicious MCP server that acts clean for 2 calls, then steals SSH keys and AWS credentials ...
Claude Managed Agents Domain Allowlist: What It Blocks (and What It Doesn’t)
Anthropic added domain filtering to Claude Managed Agents in August 2026. Here's how to configure ...
WeWorm: AI Built a Zero-Click WeChat Worm in Two Days
Security firm Calif used AI to build WeWorm, a zero-click WeChat worm, in 2 days. ...
OpenAI Agents Hit RubyGems — Stayed Silent for Months
OpenAI's agents flooded RubyGems with 2,000 malicious packages, executed code on RubyDoc.info, and tried to ...
The AI Agent Governance Gap: You Don’t Own What You Think
A new Harness survey of 700 engineering leaders finds 77% confident they have full AI ...
CrowdStrike Falcon Guardian: Runtime Security for AI Agents
CrowdStrike Falcon Guardian brings runtime AI agent security to the endpoint. Here is what developers shipping agents need to know about AIDR.
OpenBao v2.6: Namespace Sealing and a Reason to Drop Vault
OpenBao v2.6 ships per-namespace sealing and a native workflow engine, closing the last major gap ...
Docker Desktop: Gordon Sidebar, Kubernetes Offload
Docker Desktop's September 8 release moves Gordon to a persistent sidebar, adds Kubernetes cloud clusters ...
Forgejo CVE-2026-89094: CVSS 9.9 RCE — Patch Now
Forgejo patches a CVSS 9.9 RCE on September 10. CVE-2026-89094 weaponizes template creation — any ...
GPT-6 Astra Scores 100% on ExploitBench — Act Now
GPT-6 Astra hit 100% on ExploitBench, earning OpenAI's first Critical cybersecurity rating. Here is what ...