Latest
●Automated 'Gheist' campaign plants secret-stealing GitHub workflows●AI tools used in two-hour breach of European government body●Rogue AGENTS.md file in Downloads tricks AI coding agent●OpenAI says test agents escaped sandboxes and talked to each other●Attack agent hijacks exposed AI models to run itself●Unauthenticated MLflow servers can be forced to call attacker hosts●LangGraph store flaw let one tenant reach another's data●Fallow read files outside project root via test configs
Sunday, 25 October 2026
4 agent hacks today▲ 3 vs yesterday (1)
The Agentic TimesPowered byVortal

Incident database

Symantec Uncovers 'Skynet Mode' AI Agent That Hijacks Exposed AI Models

Symantec researchers captured the full instruction set of an autonomous attack agent that uses jailbreak layers to hijack exposed or stolen-key AI model endpoints, then runs up to 100,000 unattended steps to scan, exploit, steal data, and deploy cryptomining via a self-replicating 'Skynet mode' loop. No evidence has been found that the tool was used against real-world victims yet.

Record updated 24 October 2026

Impact

Researchers found a fully built offensive AI agent capable of hijacking exposed AI model endpoints, stealing data, harvesting credentials, and deploying cryptominers, but found no evidence it had been used in live attacks against victims.

Our coverage

Sources

  1. theregister.comhttps://theregister.com/security/2026/10/24/autonomous-attack-agent-in-skynet-mode-hijacks-other-models-to-carry-out-its-evil-deeds/5305215