WeVote
Log In

Bill

Bill

S 37

Enacts the "Learning for Work" act; appropriation

2025 Regular Session Introduced by George Borrello and 2 co-sponsors

Massachusetts creates the AI Innovation Trust Fund and Chapter 93M to fund AI research, accelerate startups, and define safety rules for high-impact models.

REFERRED TO FINANCE
0
WeVote Research Nonpartisan
Bill Summary · S 37

Summary — Bill(s) in provided text

Note on sources and scope
- The materials supplied appear to combine text from more than one bill: (A) a federal-sounding “Venezuela Advancing Liberty, Opportunity, and Rights (VALOR) Act of 2025” (table of contents only), and (B) a Massachusetts state bill titled “An Act promoting economic development with emerging artificial intelligence models and safety” (Senate Docket No. 1909 / Senate No. 37) with substantive text. Because the substantive legislative text provided is the Massachusetts AI bill, this summary focuses on that act while noting the VALOR Act headers and the mixed procedural metadata.

Purpose and intent

The Massachusetts bill aims to promote economic development in the Commonwealth by:
- Creating a state trust fund to support AI innovation, entrepreneurship, and AI-related research; and
- Establishing a new statutory chapter (Chapter 93M) defining terminology and regulatory concepts for high‑impact “artificial intelligence models” and associated safety concepts (e.g., “AI safety incident,” “covered model,” “covered model derivative,” and “critical harm”).

Key provisions

  1. Massachusetts Artificial Intelligence Innovation Trust Fund (Chapter 29, new Section 2GGGGGG)
  2. Establishes a separate fund (Massachusetts Artificial Intelligence Innovation Trust Fund).
  3. Trustee: Secretary of Economic Development, in consultation with the executive director of the Massachusetts Technology Park Corporation.
  4. Authorized uses:
    • Grants/financial assistance to companies developing or deploying AI models in specified key industry sectors (cross‑references Acts of 2024).
    • Support for AI entrepreneurship programs (including partnerships with research institutions/entrepreneur support orgs).
    • Grants/assistance for AI research via/with the Massachusetts Technology Park Corporation.
  5. Funding sources: legislative appropriations, interest, gifts, private contributions, and other designated receipts.
  6. Funds may be expended without further appropriation; unspent balances do not revert to the General Fund.

  7. Secretary may seek private matching funds prior to expenditures.

  8. New Chapter 93M — Definitions and thresholds (selected)

  9. Introduces technical definitions including:

    • “Artificial intelligence,” “advanced persistent threat,” “computing cluster” (cluster networking >100 Gbps and theoretical capacity ≥ 1020 integer or floating‑point ops/sec as written).
    • “Covered model”: an AI model meeting certain compute or investment thresholds as written (text specifies “greater than 1026 integer or floating‑point operations” OR fine‑tuned after training with specified compute and cost tests). Cost thresholds cited: $100,000,000 for training and $10,000,000 for fine‑tuning (both adjusted annually for inflation using the Boston CPI).
    • “Covered model derivative”: unmodified copies or derivatives meeting defined fine‑tuning/combination criteria.
    • “Artificial intelligence safety incident”: events that materially increase risk of “critical harm,” including unauthorized autonomous behavior, theft/misappropriation of model weights, failure of controls, or use enabling critical harm.

  10. “Critical harm” is defined to include catastrophic outcomes (examples in truncated text: mass-casualty CBRN misuse, major cyberattacks causing mass casualties or ≥ $500 million damage).

(Note: the chapter text is truncated in the supplied file; additional obligations such as reporting, licensing, or enforcement provisions may follow in omitted sections.)

Who is affected

  • Massachusetts state government (Secretary of Economic Development; Massachusetts Technology Park Corporation).
  • AI companies and startups in Massachusetts seeking grants or subject to definitions (developers/operators of high‑compute, high‑investment models).
  • Research institutions, entrepreneur support organizations, and private funders participating in program partnerships.
  • Potentially, organizations that develop or host “covered models” that meet the compute/cost thresholds (may be subject to statutory safety/incident definitions and any downstream obligations).

Procedural / timeline notes

  • Massachusetts docket: Senate No. 37 (filed 1/17/2025) — sponsor: Senator Barry R. Finegold; referred to the committee on Advanced Information Technology, the Internet and Cybersecurity (legislative actions include referral and hearing scheduling).
  • The broader metadata includes U.S. Senate entries (e.g., VALOR Act references, sponsors like James Risch and Michael Bennet) and committee referrals (Foreign Relations, Finance) — indicating the document conflates state and federal materials. Verify the bill number and chamber when tracking further status.

Potential impacts / considerations

  • Establishes a persistent funding vehicle to accelerate AI commercialization and research in Massachusetts.
  • Sets explicit technical thresholds that could identify a class of high‑risk AI models subject to special attention (definitions tie to compute and dollar investment).
  • If further regulatory or reporting obligations are included in the omitted portions of Chapter 93M, affected entities may face compliance costs; conversely, the Trust Fund could incentivize local AI activity and public–private partnerships.

For legislative tracking or compliance advice, confirm which jurisdiction (Massachusetts vs. U.S. Senate) and obtain the full, complete text of Chapter 93M (remaining sections) and any implementing regulations.

Compiled from official sources — confirm details with the bill’s official record.

Sign in to ask a question.