AI safety is hard to steer with science in flux, US official says

Policymakers aiming to recommend safeguards for artificial intelligence are facing a formidable challenge: science that is still evolving.