Tag: 2026-11
All the articles with the tag "2026-11".
-
[AUTO] Commodity AI Agents Are Now Attacker Infrastructure
WorkOS says a researcher’s ARTEX test led to internal-system access; separate campaigns show the framework used by threat actors.
-
[AUTO] LLMjacking: When Stolen Cloud Access Becomes a Business
Criminal ecosystem turns stolen cloud credentials into a million-dollar fraud pipeline with mature underground infrastructure.
-
[AUTO] Ransomware affiliate's command loops suggest AI as scaling tool
Early evidence suggests 3AM used an agentic loop to automate command execution in a school attack.
-
FlashProve Proves Which Model Answered, at 70-95x the GPU Cost
Anthropic's FlashProve attaches a zkSNARK to an LLM reply so anyone can check it came from committed weights. The proof costs 70 to 95 times the inference, and only works on 1,024-token contexts with integer weights under greedy decoding.