Agentjacking Attack Tricks AI Coding Agents Into Running Malicious Code

The collection highlights a surge in high-impact security events: new AI-agent attack classes (Tenet Security's "Agentjacking" via crafted Sentry error reports, OpenClaw exploits, and other inputs that coerce self‑hosted agents into running attacker code or leaking secrets); multiple high-severity/active-exploitation vulnerabilities in AI frameworks and enterprise products (LangGraph critical RCE chain, Langflow path-traversal CVE-2026-5027 under active exploitation, and Oracle PeopleSoft CVE-2026-35273 exploited by ShinyHunters); large-scale vendor patching (Microsoft fixing 206 flaws, Forti/