Researchers Demonstrate Adaptive AI Worm Using Large Language Models (heise.de)

• AI worm generates exploits dynamically via on-device LLMs
• Test network of 33 devices: 75% attacked, 65% replicating
• Preprint highlights cost shift to defenders as LLMs run on compromised hosts

"A team from University of Toronto, Vector Institute, Cambridge, and ServiceNow Research tested an AI worm that uses LLMs to generate tailor-made exploits. In isolated network tests, the worm infected 75% of 33 devices and spread to two-thirds."