HOUSE BILL REPORT
HB 1168
As Reported by House Committee On:
Technology, Economic Development, & Veterans
Title: An act relating to increasing transparency in artificial intelligence.
Brief Description: Increasing transparency in artificial intelligence.
Sponsors: Representatives Shavers, Taylor, Ryu and Fosse.
Brief History:
Committee Activity:
Technology, Economic Development, & Veterans: 1/17/25, 1/28/25 [DPS].
Brief Summary of Substitute Bill
  • Requires developers of generative artificial intelligence systems or services to post documentation regarding the data used to train the system or service.
  • Provides for enforcement by the Attorney General and establishes new civil penalties.
HOUSE COMMITTEE ON TECHNOLOGY, ECONOMIC DEVELOPMENT, & VETERANS
Majority Report: The substitute bill be substituted therefor and the substitute bill do pass.Signed by 8 members:Representatives Ryu, Chair; Kloba, Vice Chair; Cortes, Donaghy, Paul, Shavers, Simmons and Thomas.
Minority Report: Do not pass.Signed by 5 members:Representatives Barnard, Ranking Minority Member; Keaton, Penner, Volz and Waters.
Staff: Emily Poole (786-7106).
Background:

Artificial intelligence (AI) is defined and understood in different ways.  The National Artificial Intelligence Initiative Act of 2020 (Act) defines AI as a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments.  The Act further describes AI systems as systems using machine- and human-based inputs to:  (1) perceive real and virtual environments; (2) abstract such perceptions into models through analysis in an automated manner; and (3) use model inference to formulate options for information or action.

 

By Executive Order, the White House adopted the Act's definition of AI and defined related concepts, including defining "generative AI" as the class of AI models that emulate the structure and characteristics of input data in order to generate derived synthetic content, which may include images, videos, audio, text, and other digital content.  
Summary of Substitute Bill:

Required Documentation.

Before a generative AI system or service, or a substantial modification to a generative AI system or service, is made publicly available for use, the developer of the system or service is required to post documentation regarding the data used to train the generative AI system or service. 

 

The required documentation must include a high-level summary of the datasets used in the development of the generative AI system or service, including:

  • the sources or owners of the datasets;
  • a description of how the datasets further the intended purpose of the generative AI system or service;
  • the number of data points included in the datasets;
  • a description of the types of data points within the datasets;
  • whether the datasets were purchased or licensed by the developer or if the datasets were publicly available;
  • whether the datasets include personal information;
  • whether the datasets include aggregate consumer information;
  • whether there were any modifications to the datasets by the developer;
  • the dates the datasets were first trained or the date of the last significant update to the datasets during the development of the generative AI system or service; and
  • whether the generative AI system or service used or continuously uses synthetic data generation in its development.

 

Developers must comply with this requirement by January 1, 2026, for generative AI systems or services, or substantial modifications of such systems or services, that are released on or after January 1, 2022.

 

A developer is not required to post documentation regarding the data used to train a generative AI system or service that:

  • has the sole purpose of helping to ensure security and integrity;
  • has the sole purpose of operating aircraft in the national airspace; or
  • is developed for national security, military, or defense purposes and that is made available only to a federal entity.

 

Enforcement.

The Attorney General is required to enforce these requirements, and a developer who is found in violation is liable for a civil penalty of $5,000 per day of being in violation, to be collected in a civil action filed by the Attorney General.  Before bringing a civil action, the Attorney General is required to notify a developer of the alleged violation if the Attorney General determines that a cure is possible.  If the developer fails to cure the violation within 45 days after receiving notice, the Attorney General may bring a civil action without further notice.

 

Definitions.

"Developer" is defined as "a person, partnership, state or local government agency, or corporation primarily engaged in designing, coding, producing, or substantially modifying an artificial intelligence system or service intended for commercial distribution or public use."  For the purposes of this definition, "members of the public" does not include an affiliate or a hospital's medical staff member.  This definition excludes individuals or entities that develop artificial intelligence systems or services solely for internal use or research purposes or those that utilize third-party artificial intelligence systems via application programming interface without substantial modification.

 

"Generative artificial intelligence system or service" is defined as "an artificial intelligence system or service that can generate derived synthetic content, such as text, images, video, and audio, that emulates the structure and characteristics of the artificial intelligence system's or service's training data."

 

The following additional terms are defined:  aggregate consumer information; artificial intelligence; security and integrity; substantially modifies; synthetic data generation; train a generative artificial intelligence system or service; and types of data points.
Substitute Bill Compared to Original Bill:

The substitute bill:

  • specifies that "aggregate consumer information" does not mean individual consumer records that have been deidentified;
  • changes the definitions of "artificial intelligence" and "generative artificial intelligence" to definitions of "artificial intelligence system or service" and "generative artificial intelligence system or service";
  • modifies the definition of "developer" to include people or entities that are primarily engaged in designing, coding, producing, or substantially modifying an AI system or service intended for commercial distribution or public use, instead of people or entities that design, code, produce, or substantially modify an AI system or service for use by members of the public;
  • specifies that the definition of "developer" does not include individuals or entities that develop AI systems or services solely for internal use or research purposes or those that utilize third-party AI systems via application programming interface without substantial modification;
  • modifies the definition of "synthetic data generation" to refer to original data, instead of seed data;
  • replaces references to artificial intelligence with references to generative artificial intelligence;
  • removes the requirements that summaries of training data describe whether the datasets include any data protected by copyright, trademark, or patent, and the time period during which data in the datasets were collected;
  • requires the summaries of training data to include the dates the datasets were first trained or last significantly updated, instead of the dates the datasets were first used;
  • modifies the definition of personal information used in the requirement that summaries of training data describe whether the datasets include personal information from the definition used in RCW 19.255.005 to the definition used in RCW 19.373.010;
  • specifies that civil penalties are in the amount of $5,000 per day of being in violation; and
  • requires the Attorney General to provide developers with an opportunity to cure alleged violations before bringing a civil action.
Appropriation: None.
Fiscal Note: Available.  New fiscal note requested on January 28, 2025.
Effective Date of Substitute Bill: The bill takes effect 90 days after adjournment of the session in which the bill is passed.
Staff Summary of Public Testimony:

(In support) Artificial Intelligence has quickly increasing capabilities, and there are accounts of AI being used in harmful ways.  People who deploy AI models know very little about them, and AI is not a very regulated medium.  This bill will result in much-needed transparency requirements and give the public better insight into the data used to train generative AI models, including whether the data includes personal information.  The idea that requiring disclosures about AI models forces companies to disclose trade secrets is a false narrative, driven by fears of liability.  Many AI models use the same datasets, but the tools vary because of the ingenuity used in building the models.  This bill is based on a California law, and the text was debated and negotiated in California.  There are people who are worried that once they get started into a career, AI may take their job.  Studies show that image generation services have had issues with racial bias and stereotyping.  This bill will help address the copyright issues associated with generative AI tools that produce data verbatim.  This bill is narrower than people think, because it does not apply to in-house applications of AI, and it only applies to generative AI.  This bill will promote industry growth and public safety.  Other states have enacted similar bills.  Washington should be a leader in the AI industry, while balancing that leadership with responsibility.

 

(Opposed) There should not be a private right of action in this bill.  Washington should wait to legislate on AI until the AI Task Force has issued recommendations.  California has a full-time legislature, so they can continue to work on bills throughout the year.  Stakeholders, including the AI Task Force, were not given enough time to weigh in on the bill draft. 

 

(Other) A health marketplace is important for innovation.  Some of the language in the bill may result in unclear expectations.  Training data for AI models is like a nutritional label on food; it is not a recipe, but it gives information in a context that people can understand.  The bill should be amended to not encourage spurious litigation.
Persons Testifying:

(In support) Representative Clyde Shavers, prime sponsor; Jai Jaisimha, Transparency Coalition.ai; Zachary McKay, Lake Washington High School; Mihai-Alexandru Radu; and Nathan Guan.

(Opposed) Morgan Irwin, Association of Washington Business; Kelly Fukai, Washington Technology Industry Association; Rose Feliciano, TECHNET; Kelly Fukai, Washington Technology Industry Association; and Crystal Leatherman, Washington Retail Association .
(Other) Robert Singleton, Chamber of Progress.
Persons Signed In To Testify But Not Testifying: None.