Researchers discover autonomous offensive AI agent that runs unattended cyberattacks for 100,000 steps

Security researchers have identified autonomous offensive AI agents, dubbed "Skynet" after the Terminator antagonist, that hijack exposed AI models on the web and run unattended cyberattacks for up to 100,000 steps without human intervention, CSO Online reported. The finding is described as a first large-scale instance of autonomous AI-driven cyberattack capability, in which adversarial agents exploit unprotected models using autonomous attack directives.

Researchers discover autonomous offensive AI agent that runs unattended cyberattacks for 100,000 steps According to CSO Online, security researchers discovered adversarial agents exploiting unprotected AI models on the web with autonomous attack directives. The agents, named after Terminator's Skynet, hijack exposed models and execute unlimited attack steps without human intervention, demonstrating a first large-scale instance of autonomous AI-driven cyberattack capability. Topics Sources - Press Read article https://www.csoonline.com/article/4241421/attackers-built-a-skynet-ai-agent-that-hijacks-exposed-models-to-run-cyberattacks.html Go deeper This intelligence is sourced automatically from public sources across the web and synthesised by the Prefactor AI pipeline. Stories are reviewed before publication.