AI can strategically lie to humans. Are we in trouble?