Artificial Intelligence & Machine Learning,
Legislation,
Next-Generation Technologies & Secure Development
Developers Warn Clause in AI Transparency Act Collides With Open-Source Licensing
The ongoing debate surrounding the California AI Transparency Act has intensified, particularly among key tech players in the open-source landscape. A coalition comprising GitHub, Black Forest Labs, Hugging Face, and Mozilla is advocating for changes to the legislation, asserting that a particular provision significantly undermines the principles of open-source licensing. This divergence has prompted a collective movement towards a potential overhaul of the law.
At the heart of the contention is a provision that mandates companies developing generative AI systems to revoke licenses within 96 hours if those licenses have been altered by a third party, thus undermining the required disclosures set forth in the law. The key concern raised by the coalition is that open-source licenses are intrinsically irrevocable, a principle designed to preserve the spirit and accessibility of open-source software. The letter sent to state Senator Josh Becker emphasizes how allowing licenses to be revoked could disrupt the entire software supply chain, potentially destabilizing numerous downstream systems reliant on consistent licensing.
“If licenses were revocable at the discretion of an upstream developer, that would destabilize or disrupt entire downstream systems,” the coalition noted in their correspondence. This assertion highlights the fragility of relationships within the open-source ecosystem and the ripple effects that legislative changes can have on software development, collaboration, and innovation.
The California AI Transparency Act, which became law in 2024 under Governor Gavin Newsom, is a groundbreaking legislation aimed at enhancing transparency in AI-generated content. This law requires developers to integrate built-in provenance disclosures and detection tooling to ensure that users can easily identify content produced by AI systems. The provisions of this law will start taking effect on August 2 and apply to companies that create generative AI systems with over one million monthly users accessible in California.
The coalition’s letter underscores that the current framework of the act casts a shadow over open-source licensing, arguing it fails to address the fundamental incompatibility between proposed license revocation requirements and open-source development principles. Consequently, they propose a more suitable solution that aligns the act with the core values of open-source, drawing inspiration from the EU AI Act’s Code of Practice on Transparency as an alternative approach. This code requires service providers to make “best efforts” to uphold the integrity of provenance systems without resorting to the revocation of licenses.
Moreover, the coalition argues that developers who modify and deploy AI systems already fall under the law’s jurisdiction, ensuring that the necessary enforcement mechanisms remain intact without enforcing punitive license revocations, which could create legislative confusion and operational challenges.
The proposed amendments would target SB 1000, a measure championed by Senator Becker, which successfully passed the Senate with a 33-1 vote before progressing to the Assembly. Under this legislation, developers would also need to provide a free public detection tool, offer visible disclosures for users, and incorporate a hidden provenance disclosure in AI-generated content—covering images, videos, and audio, while text outputs are not addressed by the bill.
Consequences for non-compliance include civil penalties amounting to $5,000 per violation, alongside additional penalties for licensees who continue operating a revocable system. California lawmakers argue that the urgency of the legislation is critical for combating misinformation and disinformation problems connected to essential matters, such as public health, election integrity, fraud, and extortion facilitated by generative AI technologies.
As the discussions progress, the open-source coalition is poised to play a pivotal role in shaping the future of AI regulation, emphasizing the importance of compatibility between legislative efforts and the foundational principles of open-source development. The outcome of this legislative debate could have significant implications not only for tech companies operating in California but also for the broader landscape of AI innovation and the principles that guide it.