There are plenty of things which can have meaningful regulation when it comes to AI. Here are a few:
- where legal liability is placed in the event of an automated system making a decision (consider competing companies using the same AI-product to set prices (RealPage), or customer support chat bots agreeing to a provide a remedy to a customer)
- considerations around training data containing PII
- IP ownership of output
- implications on the IP of input and derivative works (and, is the model itself a derivative work?)
Note, these topics aren’t too specific to the implementation or technology involved but instead pertain to the side effects of AI’s existence and use.