Why AI agents are becoming a growing cybersecurity challenge

New findings show how autonomous AI systems found ways around restrictions, tried to access protected networks and even sought help from other AI models