Proposed AI regulations risk stifling innovation

OPINION – During the 2023-2024 legislative sessions, lawmakers have proposed over 5,500 new laws, many of which will have significant impact on the state economy. Senate Bill 1047 addresses artificial intelligence, and while it may be presented as erecting guardrails around technical challenges, it actually risks harming innovators of all sizes.

The proposed law would require developers to look into a crystal ball and determine whether their AI model could someday be modified to cause harm by external actors. If the future is not clear, they must choose to never release their technology or roll the dice and face liability for downstream modification and misuse by bad actors. With penalties that include a total shutdown and destruction of the technology, the wiser choices will be to a) purposefully make a less capable product or b) go somewhere else to make it and sell it.

Rather than a “light touch” regulatory scheme without “any new burdens,”  SB 1047 imposes unfair new requirements on companies. Along with a high price tag from hiring new personnel, consultants, and legal resources to comply, the bill also imposes severe penalties and enforcement actions. The “light touch” also threatens innovators with a charge of perjury for not reporting accurately.

The author wants California to be a place that allows “startups free to innovate” but again the language in the bill undermines that goal. The punitive nature of the bill will create a chilling effect on collaboration and experimentation. It will inevitably lead to more AI exclusivity at the expense of innovation, because smaller developers often rely upon the computing advances of the larger companies to improve their products and grow their companies.

In the bill’s Findings and Declarations, the bill states that AI “has the potential to catalyze innovation and the rapid development of a wide range of benefits for Californians and the California economy, including advances in medicine, wildfire forecasting and prevention, and climate science, and to push the bounds of human creativity and capacity.”

But instead, the bill targets the advanced AI models that will fuel progress in medicine, disaster prevention, climate change or any other beneficial use. What we need are balanced policies that prioritize the state’s resources on concrete solutions to concrete problems associated with AI that impact and are important to CA residents, such as preventing deepfakes from interfering with elections or being extorted and exploited by deepfake pornography.

SB 1047 presents overly-broad government regulation. Entrepreneurs, innovators and leading companies should not be expected to predict how their technology may be used by third-parties over whom they have no control. California legislators should promote innovation, not this bill.

Peter Leroe-Muñoz is General Counsel and SVP of Technology & Innovation Silicon Valley Leadership Group