H.R. 10044, AI Tax and Work Protection Act (bill as introduced)

H.R. 10044, introduced by Casar with Foushee and Jacobs, would tax makers and sellers of large AI models the greater of 2% of the value of tokens processed or 3% of AI-service revenue, rising as U-4 unemployment passes 5%. A new Labor Department agency would award grants for jobs in fields like child care and schools. A proposal, not law.

Why it matters

The same thing, for different readers.

What it means for you

A proposal only: it would charge the companies behind large AI models and spend the money on publicly funded jobs such as child care, elder care and local news.

For lawmakers

The tax rate would depend on the highest yearly U-4 rate of the previous three years. Grants carry wage and benefit rules; the text has no findings or revenue estimate.

For researchers

Section 405 would have the Bureau of Labor Statistics report on AI's effects on work, including displacement and lower pay, fewer hours or temporary jobs; $20 million a year authorized.

The question it raises

Grant-funded jobs must have their primary duties carried out by a natural person, a legal test of what counts as a human job, across 17 listed areas.

In their own words

Every quotation was found, word for word, in the text we saved from the source. That shows the words are there; it does not make them true, or prove the speaker meant them as quoted.

From the source’s own text.

“To impose a tax on artificial intelligence token usage and establish a Work Protection Administration within the Department of Labor, and for other purposes.”
Greg Casar, RepresentativeThe bill's own statement of purpose, as introduced by Casar with Foushee and Jacobs. A proposal, not enacted.
Found word for word in the source
“the use of a foundation model by the taxpayer or a sale, license, or exchange of such use or access to a foundation model to a related party if such use enables or results in a reduction in the workforce of the taxpayer or of such related party.”
Greg Casar, RepresentativeProposed section 4491(b)(2)(A)(ii): the second kind of taxable transaction, a company's own or related-party use that reduces its workforce.
Found word for word in the source
“Each eligible entity awarded a grant under section 401 shall use such funds to hire employees for jobs for which the primary duties are to be carried out by a natural person”
Greg Casar, RepresentativeSection 402(a): the condition on the jobs the grants would fund.
Found word for word in the source

The record

Questions
What happens to work, income and the economy? · Who can demand evidence, delay a release or stop an unsafe deployment?
People
Greg Casar (Wrote it, Representative) · Valerie Foushee (Wrote it, Representative) · Sara Jacobs (Wrote it, Representative)
What we read
We read the whole source.
Found by
A research pass: work, income and the economy, gathered October 13, 2026, via research-work (tier 1: the source’s own channel or an official record)
Cite this
US House of Representatives, “H.R. 10044, AI Tax and Work Protection Act (bill as introduced)”, August 6, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/us-house-2026-08-06-ai-tax-work-protection-act
Our notesFull text of the bill as introduced, from GovInfo. Introduced 6 August 2026 and referred to the House Education and Workforce Committee and, in addition, Ways and Means; the introduced text cannot show later cosponsors or committee action. As written, the bill would tax a person who develops a foundation model, sells access to one, or modifies an open-weight one, and who earns revenue from covered transactions or uses a model to cut its own workforce. A foundation model is a broadly trained, generally self-supervised model trained with at least 10^25 operations (printed 10\25\ in this rendering), or the amount the Secretary sets for comparable capability. Covered transactions are giving an unrelated party access to a model, or a company's own or related-party use that enables or results in a workforce reduction; research use by governments, universities, federally funded labs and 501(c)(3) bodies is excluded. The tax is the greater of a token measure (the value of tokens processed, times a rate that starts at 2 percent) and a revenue measure (payments for AI services plus the value of covered related-party transactions, times a rate that starts at 3 percent). It applies to transactions more than a year after enactment. All of it would go to a trust fund from fiscal 2027 that pays for the grants; up to 20 percent of the fund may also be authorized for Workforce Innovation and Opportunity Act programs. The rates depend on the highest annual U-4 rate of the three previous years, and the Secretary may set aside unemployment caused by a war, pandemic or other shock unrelated to AI. Read literally, the rate steps up at 7 percent unemployment: the token rate is 4 percent at 7.0 and about 6 percent just above. Grant jobs must pay at least the highest applicable minimum or prevailing wage, carry health cover comparable to federal employees', 12 weeks of paid family and medical leave and paid leave that builds with hours worked, and may not displace existing staff. Grantees are states, localities, tribes, public schools, universities and nonprofits. To check: the text gives only surnames with honorifics, so first names come from the triage record (Casar and Jacobs match existing Base people; Foushee is new), and the title Representative is inferred from the House setting. The text calls the new body both Work Protection Administration and Office of Job Creation, and cites section 4491(c) for a definition that sits in (b)(2). Section 405 asks the Bureau of Labor Statistics to report on AI's effects on the workforce beyond job displacement and job degradation; the wording is ambiguous. The quotations are the bill's text, which the three named members introduced, not words Casar wrote alone.