California lawmakers are reconsidering A.B. 412, a bill that mandates AI developers to identify and disclose all registered copyrighted works used in training generative AI models—a requirement experts say is impossible to meet and risks reinforcing the dominance of large AI companies.
- AB 412 requires impossible copyright disclosure from AI developers
- No comprehensive, accessible list of copyrighted works exists
- Bill may entrench AI industry dominance and hinder startups
What happened
California legislators are once again debating A.B. 412, which imposes new rules on AI developers to track and disclose copyrighted materials used during training of generative AI systems. The bill intends to bring transparency about the copyrighted content feeding AI models but runs into practical barriers tied to how copyright registrations and online content are managed.
EFF and other critics submitted opposition letters warning the California Senate Privacy Committee that the law’s requirements are unfeasible for developers to fulfill. Since no centralized, machine-readable database of copyrighted works exists and many copyright holders do not publicly disclose their materials, generating a complete and accurate list is nearly impossible.
Why it matters
The proposed law places a disproportionate burden on smaller AI startups, independent developers, and nonprofit or open-source projects, which typically lack the legal and compliance infrastructure of large corporations. Consequently, these entities face a high risk of being pushed out of the California market, potentially decreasing innovation and competition in AI development within the state.
Supporters of the bill claim it aims to protect creative workers’ rights against Big Tech, but critics argue that existing federal copyright law and ongoing court cases addressing AI training uses already provide remedies for rights holders. Adding a complex state-level mandate risks legal fragmentation and uncertainty without clear benefits.
What to watch next
California lawmakers’ decision on whether to pass A.B. 412 will be closely monitored by AI developers, digital rights advocates, and the tech industry at large. If enacted, the law could influence how other states approach regulation of AI training data and copyright transparency efforts.
Meanwhile, ongoing federal court rulings concerning copyright and AI training practices will also shape the legal environment. Stakeholders should track these developments, as they may provide clearer standards and reduce the need for burdensome state-specific frameworks that disproportionately affect smaller players.