H.R. 10044, the AI Tax and Work Protection Act, is a bill of the 119th Congress introduced in the House of Representatives on 6 August 2026 by Representative Greg Casar (D-TX-35) for himself and Representatives Valerie Foushee and Sara Jacobs. Its long title is "To impose a tax on artificial intelligence token usage and establish a Work Protection Administration within the Department of Labor, and for other purposes." It was referred to the Committee on Education and Workforce and additionally to the Committee on Ways and Means. One version exists, Introduced in House; the policy area is recorded as Taxation (Source: congress.gov).
The bill has four titles: an excise tax (Title I), a trust fund receiving its proceeds (Title II), a new administration within the Department of Labor (Title III), and a competitive grant program that administration would run (Title IV). The policy page for the bill and its reception is AI Tax and Work Protection Act (H.R. 10044, 119th Congress).
Title I — the excise tax
Section 101 adds a new subchapter D to chapter 36 of subtitle D of the Internal Revenue Code, containing a single new section 4491, "Excise tax on foundation models." The tax is imposed on each covered person for each taxable year in an amount equal to "the applicable amount."
Who is taxed
A "covered person" is one who (A) develops a foundation model, sells access to a foundation model, or modifies an existing open-weight foundation model, and (B) either generates revenue from a covered transaction or uses the foundation model to reduce that person's own workforce. Both limbs must be met.
A "covered transaction" is either the provision of use of or access to a foundation model to an unrelated party in the course of a trade or business, or the use of a foundation model by the taxpayer — or a sale, licence or exchange of such use to a related party — where that use "enables or results in a reduction in the workforce" of the taxpayer or the related party. Related-party status is determined by the single-employer tests of IRC sections 52(a), 52(b), 414(m) and 414(o).
Research and development use is excluded where the entity is a federal, state or local government, an institution of higher education under section 101(a) of the Higher Education Act of 1965, a Federally Funded Research and Development Center as defined at 48 C.F.R. 2.101, or a section 501(c)(3) organisation exempt under section 501(a).
The two bases
The "applicable amount" is the greater of two products:
| Base | Measure | Base rate |
|---|---|---|
| Token base | Fair market value of tokens processed by the taxpayer in covered transactions during the taxable year | 2% |
| Transaction base | Consideration received for AI services in covered transactions, plus the fair market value of all covered transactions with a related party | 3% |
The Secretary must issue and annually revise regulations or guidance, including a determination — in consultation with the Secretary of Commerce — of the fair market value of a token.
The unemployment ramp
Both rates rise with an "applicable unemployment rate," defined as the U-4 measure published in the Bureau of Labor Statistics monthly Employment Situation release, taken from whichever of the three calendar years preceding the transaction had the highest such measure, disregarding any calendar year beginning before enactment.
| Applicable U-4 rate | Token percentage | Transaction percentage |
|---|---|---|
| Not above 5% | 2% | 3% |
| Above 5%, not above 7% | 2% + (rate − 5) | 3% + (rate − 5) |
| Above 7% | 2% + 2 × (rate − 5) | 3% + 2 × (rate − 5) |
Subsection (c) lets the Secretary, in consultation with the Secretary of Labor, disregard so much of an unemployment rate above 5 percent as is determined to result from "a war, pandemic, or any other massive economic shock unrelated to the use of artificial intelligence."
Definitions
A "foundation model" is an AI model trained on broad data, generally using self-supervision, applicable across a wide range of contexts, and trained using computing power "equal to or greater than 10^25 integer or floating-point operations" — or such quantity as the Secretary determines necessary to achieve comparable model capability for the calendar year. An "open-weight foundation model" is one whose trained parameters "are made publicly available for others to download and use." A "token" is "a discrete unit of data, such as text, code, image, audio, or video data, that an artificial intelligence model processes, used for the purpose of measuring the volume of model input or output." "Artificial intelligence" takes its meaning from section 5002 of the National Artificial Intelligence Initiative Act of 2020.
The amendments apply to covered transactions occurring after the date one year after enactment.
Title II — trust fund
Section 201 establishes a Treasury trust fund receiving amounts equivalent to 100 percent of the taxes imposed by section 4491, for fiscal year 2027 and each fiscal year thereafter. Section 202 restricts its use to carrying out Titles III and IV.
Title III — the Work Protection Administration
Within 90 days of enactment the Secretary of Labor must establish a Work Protection Administration (WPA) in the Department of Labor, headed by a Director, whose duty is to develop and implement the Title IV jobs program.
The bill is internally inconsistent about the body's name: Title III is headed "Office of Job Creation at Department of Labor," section 301 establishes a "Work Protection Administration," and section 407(2) defines "Director" as "the Director of the Office of Job Creation."
Title IV — the jobs program
Section 401 directs the Director to award competitive grants to eligible entities to create employment. Priority goes to entities planning permanent, full-time jobs, and — where grant funds would support functions typically performed by state or local government — to state governments and units of general local government. The Director must consider the advisory committee's recommendations and BLS reporting on AI's labour-market impacts, and must establish an interagency task force to help review applications.
Eligible entities under section 407(3) are state governments, units of general local government, tribal governments, publicly funded elementary and secondary schools, local or state educational agencies, institutions of higher education, and section 501(c) organisations exempt under 501(a).
Grant categories
Section 402(b) enumerates seventeen categories of work: child care and early childhood education; public education programs; health initiatives; elder care and disability support; housing construction and rehabilitation; services for housing-insecure and homeless individuals; community violence prevention and public safety; scientific research and innovation; roads, bridges, rail, airports, ports, public buildings, water and wastewater systems, broadband and other public infrastructure; public art, museums, archives, libraries and local history; environmental conservation and climate resilience; wildfire prevention and disaster preparedness; repair and beautification of schools, community centers and libraries; parks, playgrounds, public and tribal lands and trails; local news and journalism, including local news funds and journalism fellowships; programs and training under the Workforce Innovation and Opportunity Act; and other activities determined by the Director and the advisory committee.
Jobs funded must be ones whose primary duties are carried out by a natural person.
Labour conditions on grantees
Section 401(e) conditions eligibility on the entity maintaining policies that do not prevent employees from exercising section 7 rights under the National Labor Relations Act; that require posting of NLRA rights notices and disclosure of pay and benefits at the start of employment; and that require recognition of an exclusive representative where more than 50 percent of employees indicate that desire. A local hiring preference for the metropolitan area of the employment site is also required. Tribal governments are exempt from these requirements.
Section 402(d) sets the terms of the funded jobs: a wage no less than the greatest of the federal minimum wage, the applicable state or local minimum, or the locally prevailing wage as determined by the Secretary of Labor; health coverage comparable to the Federal Employees Health Benefits Program; at least 12 workweeks of family and medical leave per 12-month period paid at the employee's regular rate; and one hour of paid leave for every 30 hours worked, capped at 56 hours a year unless the entity sets a higher limit.
Section 402(e) bars using grant funds to hire where doing so would lay off or partially displace an existing employee, where the new hire would assume the duties of an employee on strike, or where the new hire would perform substantially similar work to a laid-off employee who has not been offered restoration. "Partial displacement" is defined as any reduction in hours, wages or employment benefits.
Section 402(c) restricts eligibility to individuals 18 or older hired as employees, applying an ABC-style test that excludes those free from control and direction, performing work outside the entity's usual course of business, or customarily engaged in an independent trade of the same nature.
Reporting, advisory committee, and BLS
Section 403 requires an annual report to the House Education and Workforce Committee and the Senate HELP Committee on the number of grants, jobs created per grant, and the average, median and range of wages paid.
Section 404 establishes an advisory committee within 180 days of enactment (or 90 days after the office is established, whichever is sooner), meeting quarterly. Its membership comprises designees of the Secretaries of Labor, Commerce and Education, of the Chair of the Federal Reserve and of the Chair of the National Labor Relations Board; representatives of the National Science Foundation, the Council of Economic Advisers and the Office of Science and Technology Policy; at least one representative of any union with two or more branches in the United States and more than 50,000 members; the governor of each state or a designee; and two economists studying macroeconomic labour trends. The Federal Advisory Committee Act does not apply to it.
Section 405 directs the Bureau of Labor Statistics to collect and report on AI's workforce impacts "beyond" job displacement, expressly including "the degradation of existing jobs, including employees receiving lower pay or fewer hours or such jobs becoming temporary as a result of artificial intelligence," and authorises $20 million per fiscal year for FY2027 through FY2031.
Section 406 authorises appropriating not more than 20 percent of the trust fund each year to Workforce Innovation and Opportunity Act programs and training.
Textual notes
The introduced text carries several evident drafting slips, reproduced faithfully in the vault's copy: section 4491(d) reads "if such sons are treated as a single employer"; section 407(4) reads "calender month"; section 101(c) cross-references "section 4491(c)" where the definitions sit at 4491(b); and section 407(3) reads "The term an 'entity eligible' means." The naming inconsistency between the Work Protection Administration and the Office of Job Creation, noted above, is likewise in the enrolled text as introduced.
Provenance
Text as introduced, retrieved from congress.gov. The bill overview records the sponsor, the 6 August 2026 introduction, and the dual committee referral; the sponsor's office issued a press release on 7 August 2026 (Source: casar.house.gov).
Relationships
- supports: AI Tax and Work Protection Act (H.R. 10044, 119th Congress) — the primary text behind the policy page.
- related: Greg Casar — sponsor.
- related: Compute Thresholds — the 10^25 operations threshold in the definition of "foundation model."
- related: AI Labor Disruption, Inference Economics and Token Pricing, Universal Basic Income (in AI policy), Open-Weight Frontier Models