Cybersecurity briefing — 2026-04-29

A pre-authentication SQL injection in LiteLLM is being actively exploited, and attackers are going straight for the keys behind every AI gateway it fronts.

• CVE-2026-42208 SQL Injection Exploited in LiteLLM
• Vect 2.0 Ransomware Bug Causes Destructive Wiper Behavior
• Critical RCE in GitHub (CVE-2026-3854) exposed millions of repositories
• Medtronic Breach Exposes Corporate Data, Not Products
• Sleeper Extensions Turn Open VSX Into Malware Delivery
• LeRobot Deserialization Flaw Can Take Over Robots and Controllers
• Stolen Data Forums Turn Small Breaches Into Bigger Risk
• Ransomware Rivals Expose Each Other’s Operations
• State Privacy Fines Turn Into Multi-State Enforcement
• North Korean Actors Use AI to Insert Malicious npm '@validate-sdk/v2'
• Bluetooth Firmware Flaws Put Bikes and Scooters at Risk
• Marimo Zero-Day Turns AI Demo Hosting Into Malware Delivery