Geoffrey Hinton warned U.S. lawmakers they may have only about a year to impose meaningful AI guardrails before advanced systems become “out of control,” saying AI is now designing better AI and could get harder to control fast.