California's New Approach to AI Regulation: Transparency and Independent Scrutiny Take Center Stage
Remember last September when everyone was watching California's Senate Bill 1047, hoping it would become law? Well, it didn't. Governor Gavin Newsom vetoed the bill, which aimed to make AI companies test large AI models (costing $100 million or more to train) for potential dangers. While AI whistleblowers were disappointed, most big tech companies were relieved.
However, that wasn't the end of the story. Newsom, feeling the bill was too rigid, asked a group of AI researchers to come up with a better plan. He wanted something that would both encourage AI development and manage its risks. And now, their report is out.
The 52-page "California Report on Frontier Policy" points out that AI has gotten way better since Newsom vetoed SB 1047. AI can now do things like "reason" in a chain-of-thought manner. Because of this, the researchers suggest a new approach that focuses on transparency and independent reviews of AI models. Think of it as a way to keep a closer eye on AI's progress and potential problems.
Source: The Verge