California lawmakers recently concluded the 2024 legislative session, sending several artificial intelligence (AI)-related bills to Gov. Gavin Newsom for his consideration. California’s flurry of activity follows recent legislative developments in Colorado, which enacted the first state law on high-risk uses of AI earlier this year and highlights the willingness of state lawmakers to proactively direct AI policy in the US in the absence of federal legislation.
As the leading voice for the global software industry, BSA | The Software Alliance shares California lawmakers’ goal of promoting the responsible, safe development, and use of AI. BSA engaged with California’s legislators throughout the session to highlight policy approaches that can support the responsible development and use of AI – and how their proposed policies would work in practice. After careful consideration by the legislature, several closely watched AI reforms failed to advance despite significant amendments: AB 2930, which would have regulated high-risk uses of AI, and AB 3211, a proposal to address the provenance of digital content.
Among the AI bills on Gov. Newsom’s desk are several that could have unintended consequences and present practical challenges for both consumers and companies:
- Senate Bill 1047, intended to regulate frontier models of AI, is one such proposal. Earlier this year, BSA warned that, while well-intended, SB 1047 would impose a complex and unworkable regulatory approach that could discourage innovation in the Golden State. If signed into law, SB 1047 would blur the roles and responsibilities of different actors in the AI value chain in ways that will not improve consumer safety or security.
- AB 2013, a bill which risks the disclosure of trade secrets, has the potential to undermine how companies combat risk, as BSA noted earlier this year. If approved by Gov. Newsom, this reform would essentially require companies that develop or substantially modify generative AI systems and services to post detailed documentation about how their systems are trained, potentially requiring companies to disclose confidential information. While intended to improve transparency, this approach could disincentivize California companies from training systems on a broader set of data, which helps to both improve the performance of AI systems and protect against bias. It would also impose reporting requirements that go well beyond current laws, including those in the European Union.
- AB 1008 amends the definition of personal information under the California Consumer Privacy Act. This change risks expanding the scope of the state’s consumer privacy law in a manner that could create confusion for consumers looking to understand their rights and for companies looking to understand their obligations.
In addition to these measures, California lawmakers considered dozens of other AI-related reforms. The legislative session ended with many outstanding AI-related policy questions that are likely to be the focus of ongoing debate and discussion among policymakers. Many of these discussions will continue into the 2025 session, with similar trends playing out across the country.
BSA thanks Gov. Newsom for his careful consideration of the reforms now before him and commends the California legislature for its willingness to engage with a diversity of stakeholders on these critical issues. BSA looks forward to continued dialogue on behalf of the enterprise software industry in California and across the country.
