Sacramento –
Governor Gavin Newsom recently signed a bill aimed at enhancing transparency for large AI companies. This new legislation mandates the disclosure of security protocols and the reporting of significant safety incidents.
Senator Scott Wiener (D-San Francisco) described Senate Bill 53 as a means to establish “Commonsense Guardrails” that prioritize both safety and transparency amid the rapidly advancing field of artificial intelligence. Newsom expressed that this bill effectively balances collaboration with AI firms while avoiding excessive regulation.
“AI is an innovative frontier, and California is not just participating; it’s establishing itself as a national leader by implementing groundbreaking AI safety laws that foster public trust as this technology quickly develops,” Newsom remarked in a statement.
This bill was introduced following Newsom’s rejection of a broader measure last year, also authored by Wiener. The previous bill, SB 1047, had notable backing from figures like Elon Musk and prominent AI researchers but faced opposition from organizations like Meta and OpenAI.
In a lengthy veto message last year, Newsom acknowledged SB 1047’s “good intentions” but critiqued its effectiveness in safeguarding the public from genuine technological threats. He then announced plans for his administration to assemble a working group comprising AI leaders and experts to create more practical protections, which eventually laid the groundwork for SB 53.
The newly enacted law obligates businesses to reveal their safety and security measures and conduct risk assessments. It also requires the state’s emergency services department to log significant incidents, including cyberattacks and unsafe actions by autonomous AI systems.
Starting in 2027, the California Office of Emergency Services (CAL OES) will release annual reports that aggregate and anonymize critical safety incident data. Additionally, SB 53 strengthens protections for whistleblowers who report violations within their companies.
Under this law, the California Attorney General can impose civil penalties reaching up to $1 million for non-compliance.
“When engaging with transformative technologies like AI, we have a duty to nurture this innovation while also establishing Commonsense Guardrails to assess and reduce risk,” Wiener stated.
The bill faced opposition from the California Chamber of Commerce and the High Tech Industry Association.
“This rigorous approach compels developers to divert crucial time and resources towards managing virtual risks rather than focusing on tangible harms,” the Progressive Room commented.
