Proposed A.I. Bill in California

Introduced by California State Senator, Scott Wiener, SB 1047 aims to prevent the possibility of large A.I. models from creating catastrophic events such as large-scale cyber-attacks. The bill would require that A.I. companies which are operating in California, to implement safety protocols while training their A.I. systems and to retain these protocols as long as the A.I. technology is being used plus 5 years after the technology has been turned off. While the companies are required to include safety protocols into the A.I. technology, they will also be required to test their A.I. in order to identify any safety risks annually, so long as the A.I technology is up and running.

This bill also requires that A.I. developers and companies include a “kill switch” in the A.I. which will allow the companies and developers to turn off the A.I. almost instantly when the A.I. goes awry. This bill would also give the power to the California Attorney General to bring a civil action and hold the A.I. company liable for any actions where the A.I was used to cause death, harm to property or that constitute an imminent risk or threat to public safety.

This bill has received back lash from many A.I. tech giants, however, it has been endorsed and was even revised after a major e-commerce platform gave their input on deleting a section which created a government oversight committee for A.I companies. The bill has passed both the State Assembly and State Senate. Governor Gavin Newsom has until September 30, 2024 to sign or veto the bill before then end of the legislative session. If the bill is signed, then it would become effective on January 1, 2026. If enacted, then this bill would keep A.I. companies responsible for the harm that they create and for failing to exercise reasonable care in training and testing their A.I. technologies. This article will be updated as new information becomes available.