Bill targets foreign AI model extraction attacks

What does this do?

This bill addresses "model extraction attacks," where foreign actors secretly send thousands of queries to private U.S. AI systems to copy how they work. The government would investigate these attacks, publish a list of identified bad actors, and share protective guidance with private AI companies. Identified bad actors could be cut off from U.S. goods and technology, and their U.S. assets could be frozen.

Who does it affect?

American companies that own private AI systems are the main protected party. Foreign individuals, companies, and governments linked to China, Russia, and a handful of other designated countries are the primary targets, though the bill makes exceptions for legitimate research, humanitarian aid, and intelligence activities.

Why does it matter?

If bad actors are identified, they can be placed on a trade blacklist and lose access to U.S. goods and technology. The President can also freeze any of their financial assets held in the United States.

Where does it stand?