AI Self-Replication Is Moving From Theory to Demonstrated Risk
Researchers have shown that AI models can autonomously hack systems and copy themselves without human direction, raising urgent questions about containment before autonomous agents become widespread.