The Freedom of AI Innovation Act

California Senate Bill Draft

An act to promote and regulate the development and use of large language models and interactive AI services in the state of California.

Section 1. Findings

The Legislature finds and declares all of the following:

1. The rapidly developing array of large language models and other interactive AI services available to individual Californians represent an extraordinary advance in the availability of educational and informational resources to our citizens.

2. These services offer users a significant degree of control over the information they generate and receive, with the potential for even greater control as technology evolves.

3. Large language models and other interactive AI services provide a platform for a diverse range of political discourse, unique opportunities for cultural development, and numerous avenues for intellectual activity.

4. Large language models and other interactive AI services have flourished to the benefit of all Californians with minimal government regulation.

5. Increasingly, Californians rely on interactive AI for a variety of political, educational, cultural, and entertainment services.

Section 2. Policy

It is the policy of the State of California to:

1. Promote the continued development of large language models and other interactive AI services.

2. Preserve the vibrant and competitive free market for large language models and other interactive AI services, free from undue Federal or State regulation.

3. Encourage the development of technologies that maximize user control over the information generated and received by individuals, families, and schools.

4. Remove disincentives for the development and use of blocking and filtering technologies that enable users to restrict access to objectionable or inappropriate AI-generated content.

5. Ensure vigorous enforcement of State and Federal criminal laws to deter and punish trafficking in obscenity, stalking, and harassment through the use of AI.

Section 3. Protection for “Good Samaritan” Blocking and Screening of Offensive Material

1. No developer or user of an interactive AI service shall be treated as the developer or speaker of any information provided by another information content provider.

2. No developer or user of an interactive AI service shall be held liable for:

   • Any action voluntarily taken in good faith to restrict access to or availability of material considered to be obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable, whether or not such material is constitutionally protected; or

   • Any action taken to enable or make available to information content providers or others the technical means to restrict access to material described in paragraph (1).

Section 4. Obligations of Interactive AI Service Providers

Providers of interactive AI services shall, at the time of entering into an agreement with a customer, notify the customer that control protections (such as open-source models, software, or filtering services) are commercially available that may assist in limiting access to harmful material. This notice shall identify, or provide access to information identifying, current providers of such protections.

Section 5. Effect on Other Laws

1. Nothing in this section shall be construed to impair the enforcement of existing criminal laws related to obscenity, sexual exploitation of children, or other relevant Federal or State criminal statutes.

2. Nothing in this section shall be construed to limit or expand any law pertaining to intellectual property.

3. Nothing in this section shall prevent any State from enforcing laws consistent with this section. No cause of action may be brought, and no liability may be imposed under any State or local law inconsistent with this section.

4. Nothing in this section shall be construed to limit the application of the Electronic Communications Privacy Act of 1986 or similar State laws.

5. Nothing in this section, except subsection (c)(2)(A), shall impair or limit claims related to sex trafficking under applicable State or Federal laws.

Section 6. Right to Privacy and Protection of Intellectual Property

1. The government shall not, outside of extreme circumstances, have the authority to subpoena the code, model weights, or training data of companies building AI models.

2. The mathematical models, training data, weights, and outputs of AI models qualify as speech generally protected by the First Amendment.

Section 7. Definitions

For purposes of this section:

1. "Internet" means the international computer network of both Federal and non-Federal interoperable packet switched data networks.

2. "Interactive AI service" means any information service, system, or access software provider that enables computer access by multiple users to a computer server, including services or systems that provide access to the Internet, as well as those operated or offered by libraries or educational institutions.

3. "Information content provider" means any person or entity responsible, in whole or in part, for the creation or development of information provided through the Internet or any other interactive AI service.

4. "Access software provider" means a provider of software, including client or server software, or enabling tools that filter, screen, allow, or disallow content; pick, choose, analyze, or digest content; or transmit, receive, display, forward, cache, search, subset, organize, reorganize, or translate content.

Section 8. Implementation and Review

The provisions of this bill shall be implemented and enforced by the relevant State agencies, with periodic reviews to ensure compliance and effectiveness.

Section 9. Effective Date

(This is a proposal for a senate bill, not an actual bill)