KPMG has laid out a series of safeguards aimed at preventing AI agents from producing unpredictable or harmful outcomes, a ...