Tech lobby and a tight budget end bid for Washington State AI transparency bill

Tech lobby pushback, as well as concerns over the cost of enforcement, killed a training data transparency bill in the House. Washington state is facing a $10-15 billion budget shortfall this session. (Image by Mohamed Hassan from Pixabay.)

Mar. 1, 2025 — A bill that would increase transparency requirements for AI developers has fallen victim to the tech lobby and a tight budget year in the Washington State legislature.

Friday, Feb. 28, was the deadline for bills to pass through fiscal committees in the House and Senate. As the Seattle Times noted earlier today: “With a multibillion-dollar deficit hanging over the session, bills creating new or expensive programs have proved to be a tough sell.”

HB 1168, authored by Rep. Clyde Shavers with input from the Transparency Coalition, was intended to reinforce the industry standards established by the 2024 California AI Training Data Transparency Act. It wasn’t expensive, but it did propose a law that required the state Office of the Attorney General to enforced it—and that represented a budget line.

Because of that, the bill needed to be approved by the House Appropriations Committee by the end of business on Friday in order to continue on in Olympia. By the end of the day, however, HB 1168 was one among many bills left unapproved by the committee.

A warm early reception

The bill was approved on Jan. 28 by the House Committee on Technology, Economic Development, & Veterans, and received a positive reception during an Appropriations Committee hearing on Feb. 25. TCAI co-founder Jai Jaisimha was among a number of others testifying on behalf of the bill. That committee hearing can be seen on video here, with consideration of HB 1168 starting at the 1:33:00 mark.

The bill was not voted on by the committee, however, as it faced pushback from local and national tech lobbyists (Microsoft and Amazon are both based in Washington) and stiff competition from other bills fighting for funding in an extremely lean year.

Screen image of Jai Jaisimha testifying before the House committee on February 25, 2025.

TCAI Co-founder Jai Jaisimha testified in favor of HB 1168 on Tuesday, Feb. 25.

Tech giants want no transparency

The provisions in HB 1168 followed closely on the language and requirements in the California Training Data Transparency Act, also known as AB 2013. That bill was adopted in Sacramento last September, and all AI developers reaching customers and end users in California will have to abide by it starting in early 2026.

Nevertheless, local tech giants Microsoft and Amazon, in step with national tech lobbyists, made their presence felt in Olympia. The corporate tech lobby has resisted most efforts to establish ethical standards, the legal licensing of AI training data, and requirements for a basic high-level data ingredients list.

Despite the efforts of bill author Shavers, who amended the original language to bring the bill closer to a form acceptable to local tech lobbyists, a few key Democrats on the Appropriations Committee shrank from supporting the measure.

many bills, tight budget

The bill also faced stiff headwinds regarding its enforcement provisions. 2025 is an unusually difficult budget year for Washington. The state writes and approves its biennial budget during odd-year legislative sessions, and this year the state revenue forecast projects a $10-$15 billion shortfall during the 2025-2027 budgetary cycle. Washington’s budgetary year runs July 1 to June 30.

That means every bill that crossed the desk of the House Appropriations Committee, with its singular focus on financial cost, received extra sharp scrutiny this year. Gov. Bob Ferguson last week proposed $4 billion in budgetary cuts and savings to start closing the expected gap.

Enforcement costs borne by the attorney general

With that in mind, the estimated cost of enforcing the requirements set out in HB 1168 appeared to be a high hurdle to overcome.

The bill’s transparency requirements would be enforced by the Office of the Attorney General, without an optional private right of action. The Attorney General’s office turned in an estimated cost of $1.6 million for enforcement in 2025-2027, $1 million in 2027-2029, and $0.5 million in 2029-2031.

That may not be much within a total biennial budget expected to cost roughly $70 billion—but in a year where every state agency is being asked to trim costs to the bone, it might have been enough to give some legislators pause.


More on ai legislation in 2025




Previous
Previous

How the growing market for training data is eroding the AI case for copyright ‘fair use’

Next
Next

TCAI’s Top Five: California AI bills to watch right now