Hinton says open weights aid cyberattacks — but admits the battle is lost
geoffrey hinton, fei-fei li, and andrew ng argue for open ai to prevent monopolistic control, despite hinton's fears that open weights lower the barrier for cyberattacks.
at ai4, andrew ng, fei-fei li, and geoffrey hinton urged the industry to maintain open ai access to avoid monopolistic control by major labs, comparing potential consolidation to apple and google's mobile dominance. hinton clarified his stance by differentiating open-source code from open weights, arguing that releasing trained model weights makes it cheaper and easier for bad actors to conduct cyberattacks. despite these risks, hinton conceded that the battle against open-weight models is lost and the barrier to accessing foundation models has permanently disappeared. ng framed the issue as a matter of geopolitical competitiveness, arguing that restricting open-source ai in the us could allow chinese models to spread unchecked across the developing world.