Regulon — VA-HB-713 · VA HB 713 (FAIR AI Act)

Summary

Virginia HB 713 (FAIR AI Act) would have imposed a limited transparency obligation on developers of base AI models, requiring them to disclose basic model information (name, developer, incorporation location, version date, training data recency, supported languages, and terms of service link) in their terms of service. The bill also eliminates 'the AI did it' as a defense in civil or criminal actions — developers cannot claim an AI system autonomously caused harm, and deployers cannot shift blame to the AI system. A FAIR AI Enforcement Fund would be established under the Attorney General's authority to support state agency enforcement of AI misuse, bias, and workforce disruption. The bill was left in committee and did not advance.

Enforcement & Penalties

Enforcement Authority

The FAIR AI Enforcement Fund is administered by the Attorney General, who authorizes disbursements for state agency enforcement of AI system misuse, bias, and workforce disruption. No private right of action is explicitly created. No specific complaint-driven or agency-initiated enforcement mechanism is detailed beyond the Fund's purpose of supporting state agency enforcement.

Penalties

No specific damages, penalties, or remedies are enumerated in the statute. The FAIR AI Enforcement Fund supports state agency enforcement but the statute does not specify civil penalty amounts, statutory damages, or injunctive relief provisions.

Who Is Covered

"Deployer" means any person doing business in the Commonwealth that deploys or uses an artificial intelligence system to make a consequential decision in the Commonwealth.

"Developer" means any person doing business in the Commonwealth that develops or modifies an artificial intelligence system that is offered, sold, leased, given, or otherwise made available to deployers or consumers in the Commonwealth.

What Is Covered

"Artificial intelligence system" means any machine learning-based system that, for any explicit or implicit objective, infers from the inputs such system receives how to generate outputs, including content, decisions, predictions, and recommendations, that can influence physical or virtual environments. "Artificial intelligence system" does not include any artificial intelligence system or base artificial intelligence model that is used for development, prototyping, and research activities before such artificial intelligence system or base artificial intelligence model is made available to deployers or consumers.

"Base artificial intelligence model" means a large-scale artificial intelligence system model trained on broad and diverse data to learn general patterns and relationships, serving as a foundational platform adaptable for various specific tasks or applications.

Compliance Obligations 4 obligations · click obligation ID to open requirement page

G-02 Public Transparency & Documentation · G-02.1 · Developer · Foundation Model

Va. Code § 59.1-615(A)(1)-(7)

Plain Language

Developers of base AI models (i.e., large-scale foundation models trained on broad data) must clearly and conspicuously disclose seven categories of basic model information in the terms of service: model name, developer identity, developer's incorporation location, most recent version release date, training data recency date, supported languages, and a link to the terms of service. The disclosure must be appropriate for the medium and easily accessible to users. This is a relatively lightweight model card-style obligation focused on identity and provenance metadata rather than capabilities, limitations, or safety assessments.

Statutory Text

A. A developer of a base artificial intelligence model shall clearly and conspicuously disclose, in a manner that is appropriate for the medium of the content and is easily accessible to the user of such model, in the terms of service governing the use of such model: 1. The name of the model; 2. The developer of the model; 3. The location where the developer is incorporated; 4. The release date of the most recent version of the model; 5. The date that the model's training data was most recently updated; 6. Supported languages for the model; and 7. A link to the model's terms of service.

Other · Foundation Model

Va. Code § 59.1-615(B)

Plain Language

Complying with the disclosure requirements in § 59.1-615(A) does not shield a developer from liability for harm caused by the base AI model. This is a liability clarification — it ensures that disclosure is not treated as a safe harbor against other claims. It creates no new affirmative compliance obligation.

Statutory Text

B. The provision of such disclosure to a user shall not be a defense to liability for any harm caused to a plaintiff.

Other · Foundation Model

Va. Code § 59.1-616

Plain Language

The FAIR AI Enforcement Fund is a special nonreverting fund in the state treasury, administered by the Attorney General, to support state agency enforcement against AI system misuse, bias, and workforce disruption. This is an administrative and appropriations provision that establishes the enforcement funding mechanism but imposes no compliance obligation on developers, deployers, or other private entities.

Statutory Text

There is hereby created in the state treasury a special nonreverting fund to be known as the FAIR AI Enforcement Fund. The Fund shall be established on the books of the Comptroller. All funds appropriated for such purpose shall be paid into the state treasury and credited to the Fund. Interest earned on moneys in the Fund shall remain in the Fund and be credited to it. Any moneys remaining in the Fund, including interest thereon, at the end of each fiscal year shall not revert to the general fund but shall remain in the Fund. Moneys in the Fund shall be used solely for the purpose of supporting state agency enforcement of artificial intelligence system misuse, bias, and workforce disruption. Expenditures and disbursements from the Fund shall be made by the State Treasurer on warrants issued by the Comptroller upon written request signed by the Attorney General.

Other · Foundation ModelAutomated Decisionmaking

Va. Code § 59.1-617(A)-(C)

Plain Language

Developers cannot defend against claims by arguing that the AI system autonomously caused the harm rather than the developer. Deployers cannot defend against claims by arguing that the AI system — rather than the deployer — caused the harm. Both provisions close the 'the AI did it' defense loophole, ensuring that developers and deployers remain accountable for harm caused by their AI systems. Existing common law defenses, including intervening or superseding cause, remain available. This provision modifies the litigation landscape but creates no new affirmative compliance obligation — it affects how liability is litigated, not what entities must do to comply.

Statutory Text

A. In any criminal or civil action against a defendant that is alleged to have developed or modified an artificial intelligence system that caused harm to a plaintiff, it shall not be a defense that the artificial intelligence system autonomously caused such harm to the plaintiff. B. In any criminal or civil action against a defendant that is alleged to have deployed or otherwise used an artificial intelligence system that caused harm to a plaintiff, it shall not be a defense that the alleged harm was caused by the artificial intelligence system. C. Nothing in this section shall prevent a defendant from asserting existing available defenses at common law, including that the harm to a plaintiff was caused by the intervening or superseding conduct of another individual.