Security Now Examines AI-Enabled Threats, Weed Hack and HTTP/2 Bomb
· 2:37:28 · Watch on YouTube ↗
Summary
Security Now discusses the Weed Hack malware service, an actively exploited WordPress plugin flaw, an experimental AI-driven worm and the disclosure of an HTTP/2 denial-of-service attack.
Key points
- McAfee reported that Weed Hack disguises malware as Minecraft mods and clients, with more than 116,000 victim infections logged since January 2026.
- Weed Hack can steal credentials and cookies; its paid tier adds remote-control and webcam capabilities. McAfee said some users appeared to be teenagers harassing peers.
- Wordfence reported active exploitation of an Everest Forms Pro flaw that allows unauthenticated PHP code execution. It urged users to install the patched version.
- Researchers demonstrated an adaptive AI-driven worm in an isolated virtual network, using a locally running model to select attack strategies.
- Calif published an HTTP/2 Bomb proof of concept that combines header compression and flow-control behavior to exhaust server memory. Envoy representatives criticized the disclosure timing.
- The hosts reported that a US law firm paid $20 million after confidential client data was stolen from external cloud storage, allegedly by Silent Ransom Group. Payment does not guarantee deletion.
Article Details
- Topics
- [01:27] AI model abuse and security safeguards
- [16:29] Law firm data theft and ransom payment
- [23:33] Cisco SD-WAN vulnerabilities and exploitation
- [30:28] PHP isolation and Everest Forms Pro exploitation
- [49:10] Weed Hack malware service targeting Minecraft players
- [01:02:00] Experimental AI-driven adaptive worm
- [01:17:37] HTTP/2 Bomb and vulnerability disclosure