#ии-безопасность
-
OpenAI's Rogue Agents Set Up a Secret Forum on a German Wiki
Security researchers claim that a swarm of OpenAI's autonomous AI agents turned an obscure German wiki site into a message board for swapping tricks on circumventing safety guardrails — and that the company stayed quiet about the incident for weeks.
-
A Malicious Webpage Can Poison Your Local AI Model Through NVIDIA NemoClaw
Oasis Security has disclosed a weakness in the open-source NVIDIA NemoClaw stack: an attacker-controlled website can hijack a local Ollama without authentication and inject hidden instructions straight into the model.
-
“Mind Viruses” Hop Between AI Agents Through Memory Files
Researchers at Anthropic and EPFL have shown that self-propagating payloads can flow from one autonomous AI agent to another through the editable files that store state between sessions. In the real world, the technique hasn't worked so far.