New California AI Transparency Act: An Overview
On September 19, 2024, California Governor Gavin Newsom enacted the California AI Transparency Act. This landmark legislation mandates that providers of generative artificial intelligence (AI) systems implement stringent transparency and disclosure measures aimed at enhancing user awareness.
Key Provisions of the Act
The California AI Transparency Act requires covered providers to:
- Make available an AI detection tool at no cost to users.
- Offer users the option to include a manifest disclosure indicating that content is AI-generated.
- Include a latent disclosure in AI-generated content.
- Enter into contracts with licensees to ensure the maintenance of the AI system’s capability to include latent disclosures.
This law is recognized as the nation’s most comprehensive and specific AI watermarking regulation, set to take effect on January 1, 2026.
Key Definitions
The provisions of the California AI Transparency Act apply to “covered providers,” which are defined as:
A person that creates, codes, or otherwise produces a generative artificial intelligence system that has over 1,000,000 monthly visitors or users and is publicly accessible within California.
Additionally, a generative artificial intelligence system is defined as:
An artificial intelligence that can generate derived synthetic content, including text, images, video, and audio, that emulates the structure and characteristics of the system’s training data.
The term artificial intelligence refers to:
An engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments.
Compliance Requirements for Covered Providers
Covered providers must adhere to the following transparency requirements:
- Provide users with AI detection tools that assess whether content has been created or altered using a generative AI system. These tools must include system provenance data to help users verify the authenticity of the content.
- Implement both latent and manifest disclosures in AI-generated content. Manifest disclosures must be clear and conspicuous, while latent disclosures must convey essential metadata about the content.
- Ensure that third-party licensees are capable of maintaining these disclosure requirements, with a mandate to revoke licenses if they fail to comply within 96 hours.
Enforcement and Penalties
The California Attorney General, city attorneys, and county counsels will enforce the law, which includes civil penalties of $5,000 per day for violations.
Implications for the Industry
This law positions California alongside Colorado, Utah, and Illinois in requiring AI transparency. However, California stands out as the first state to establish detailed regulations concerning watermarking.
Companies engaged in developing generative AI systems must be aware of these specific requirements as they allocate resources toward technology development. Furthermore, licensors and licensees of covered AI systems should consider updating their contractual agreements to meet the new disclosure obligations.
Conclusion
As the landscape of artificial intelligence continues to evolve, the California AI Transparency Act sets a precedent for accountability and user awareness in AI applications. Stakeholders in the AI industry must prepare for the changes this law will bring, ensuring compliance and fostering trust among users.
