California Passes Pioneering Law to Regulate Powerful AI Systems for Safety
-
California introduced new AI legislation focused on regulating large-scale AI systems that could cause mass harm. It could provide a model for national regulation.
-
The bill defines "frontier" AI models as substantially more powerful than existing systems, costing over $100 million to build. Requirements include access controls, ability to shut down in a safety incident, and notifying California of safeguards.
-
The bill had input from leading AI researchers and has support from experts, though some critics worry it may be toothless if unable to fully shut down dangerous systems.
-
The law addresses catastrophic risks from powerful AI but not other concerns like bias, discrimination, unemployment. Those require different solutions.
-
This is just the start of policymaking needed to ensure AI development proceeds safely and reaches its potential to benefit humanity.