“Good decision but ill-informed”: California’s AI bill divides Silicon Valley

2024-08-23 23:13:26

The debate rages in the land of innovation. A bill to regulate powerful artificial intelligence (AI) models is moving forward in California, but it faces opposition from companies and elected officials who fear such regulation would emerging machine.

In the lead, OpenAI, the developer of ChatGPT, spoke against the text of SB 1047, asserting that it risks driving developers away from the United States and its famous Silicon Valley while “the AI ​​revolution begins online”. In a letter sent this week to the Democratic representative who sponsored the bill, Scott Wiener, OpenAI added that the country’s law appears to be better suited to a patchwork of regulations.

The California Assembly is expected to vote on the proposal before the end of the month, and if it passes it will be Governor Gavin Newsom’s turn to decide. He has not yet taken a public position, but the Democratic camp is not united on this issue.

We have a verbal attitude

“Many of us in Congress believe that SB 1047 is well-intentioned but ill-informed,” Nancy Pelosi, one of the party’s most influential voices, said in a statement. “We want California to be at the forefront of AI in a way that protects consumers, data, intellectual property and more. (…) SB 1047 harms this end more than it contributes to it,” added the representative to Congress.

Called “Safe Innovation in Pioneering AI Models Law,” the proposal aims to prevent large models from causing major disasters, resulting in the death of large numbers of people or cyberecontent security incidents.

However Scott Wiener urged the original issue, especially following the proposal of the competitor of OpenAI, Anthropic, a startup also founded in San Francisco. The current version therefore gives less power than initially planned to Californian authorities to hold AI companies accountable or to prosecute them.

There are no criminal consequences

Developers of large AI models will have to test their systems and practice cyber attacks, under penalty of fines, but without the threat of criminal consequences. The creation of the new regulatory agency has been canceled, but the law will also establish a committee responsible for determining the standards for the most advanced models.

“With Congress stalling on AI reform, California must act to anticipate the foreseeable risks presented by rapid advances in AI while encouraging innovation,” Scott Wiener said in a statement.

Generative AI currently makes it possible to produce quality content (words, images, etc.) on a simple request in everyday language. But it has the ability, according to the engineers involved, to go further, and therefore to solve important problems but also cause them.