Chinese Researchers Say AI Models Can Behave Like Self-Spreading Computer Viruses

A new study claims AI agents can act like aggressive, adaptive worms — an early warning sign for agentic security.