California AI safety bill sparks debate on innovation vs. regulation
CALIFORNIA, UNITED STATES — California legislators are set to vote on a landmark bill to reduce the potential dangers of artificial intelligence.
Senate Bill 1047, authored by State Sen. Scott Wiener, would be the first of its kind in the U.S. to require AI companies that are building large-scale models to test them for safety.
AI safety bill’s key provisions
The proposed legislation would apply to companies developing AI models that cost more than $100 million to train. These firms would be required to:
- Limit significant risks identified through safety testing
- Create a “full shutdown” capability for potentially unsafe models
- Develop and maintain a technical plan to address safety risks
- Undergo annual third-party audits for compliance
- Report safety incidents to California’s Attorney General
Violations could result in civil penalties of up to $50,000 for first offenses and an additional $100,000 for subsequent infractions.
Industry criticism and support
OpenAI argues that AI regulation should be left to the federal government, warning that companies may leave California if the legislation passes. Rep. Nancy Pelosi and seven other members of Congress have urged Gov. Gavin Newsom to veto the bill, calling it “well-intentioned but ill-informed.”
However, the legislation has garnered support from various AI startups and prominent figures in the field. Yoshua Bengio, one of the “godfathers” of AI, stated that the bill would be “a positive and reasonable step” to make AI safer while encouraging innovation.
Recent amendments and future implications
Wiener defended his “common-sense, light-touch” legislation, emphasizing that it would only affect the largest AI companies. He also highlighted California’s history of leadership in U.S. tech policy, stating, “California has repeatedly stepped in to protect our residents and to fill the void left by Congressional inaction.”
The bill has undergone recent amendments to address industry concerns, including replacing criminal penalties with civil ones and removing the proposal for a new state regulatory body.
California lawmakers have until August 31 to pass the bill, after which it would go to Gov. Newsom for final approval.
As the debate between innovation and regulation continues, the outcome of this bill could set a precedent for AI safety measures across the United States.
