Developments in AI regulation, standards, and enforcement, each with a short read on why it matters for small and medium organizations.
Material U.S. federal, state, and local requirements — with legal status, key dates, official sources, and readiness actions.
Proposed — not bindingPending federal legislation on assurance, documentation, testing, liability, and human oversight. Nothing here is law.
FeaturedWhat SB 813 and AB 1405 mean for AI audits, independence, evidence retention, and qualification readiness.
California enacts four AI bills requiring human corroboration and stronger documentation; Connecticut's CART Act provisions take effect; a Treasury inspector general audit finds missing IRS AI impact assessments.
California's attorney general serves an investigative subpoena on OpenAI while New Mexico unveils proposed frontier-AI legislation that would authorize state-approved independent auditors.
Senators Hawley and Murphy announce the bipartisan AI Agent Accountability Act, which would create liability for reckless AI-agent operation and inadequate developer safeguards.
Two federal bills target AI-powered license-plate reader networks while Connecticut's enacted privacy law adds surveillance pricing, facial-recognition and geolocation provisions effective October 1.
The White House announces a Super Intelligence Force coordination initiative following a voluntary September 29 accord in which six major AI companies committed to layered assurance including independent external auditors.
Federal Regulatory Watch. On September 29, 2026 the President issued two executive orders that materially change federal AI governance. The America.gov order makes accuracy, reliability, transparency, data minimization, secure authentication and auditable authorization operative conditions for using super intelligence — the Administration's new term for AI — in federal services, with an OMB implementation memorandum due within 90 days. The companion order directs agencies to use "Super Intelligence" and "SI" in place of "Artificial Intelligence" and "AI", with proposed legislative language for a federal SI definition due within 60 days. Both bind the executive branch only; neither creates a general private-sector AI license, independent-audit mandate or certification regime, though contractors may encounter the controls through procurement and forthcoming OMB instructions.
Congressional Regulatory Watch. Between September 23 and 24, 2026, U.S. senators introduced or announced three separate AI proposals: the bipartisan AI Systems Transparency Act (FTC-enforced model disclosures), the Cybersecurity and AI Board of Investigations Act (an independent incident-investigation board), and the AI Regulator Act proposal (a Federal Digital Commission with pre-certification review of frontier models). None is binding law.
State Regulatory Watch. On September 17, 2026, the New York Public Service Commission (Case 26-M-0552) ordered 12 named gas, electric and water utilities to submit, within 60 days, an inventory of all operational AI use cases and the policies, procedures and protocols governing AI use. This is a binding order for those utilities — not a comprehensive AI rule or certification regime.
Enforcement Watch. Florida Attorney General James Uthmeier has filed a motion for a temporary injunction against OpenAI and CEO Sam Altman in Highlands County Circuit Court, asking the court to restrict new-model development without safeguards approved by an independent expert third party, among other relief. This is requested relief — not a binding AI requirement unless and until a court grants it.
A bipartisan coalition of 26 attorneys general asked Congress to establish comprehensive federal regulation and safety protocols for frontier AI. The September 23 letter is nonbinding advocacy that creates no private-sector duty, deadline or certification requirement — but its proposed structure signals what a future federal AI-assurance regime could demand.
On September 23, 2026, Governor Tina Kotek issued Executive Order 26-26, effective immediately, directing Oregon’s State Chief Information Officer to develop standards and criteria for adequate independent third-party AI safety review in the procurement and use of frontier AI models, and to assess the viability of a frontier-model “kill switch.” The order is binding direction to state government — not a private-sector regulation — and the review criteria are not yet final. An implementation proposal is due within 90 days and the Governor reassesses the order every three months.
On September 22, 2026, Governor JB Pritzker issued Executive Order 2026-07, effective immediately, establishing the Illinois Artificial Intelligence Cabinet. The Cabinet will assess AI risks, prepare the state for AI-related incidents, protect public infrastructure, and recommend additional AI-safety and accountability measures. The order binds and directs Illinois state government — it does not itself impose new requirements on private organizations.
On September 21, 2026, Governor Kathy Hochul announced the operational rollout of New York's Responsible AI Safety and Education (RAISE) Act, including a November registration and preparation phase for large frontier-AI developers and the appointment of the first full-time implementation official in the Office of Digital Innovation, Governance, Integrity and Trust (DIGIT). The substantive statutory requirements become effective January 1, 2027.
On September 22, 2026, Governor Wes Moore announced an AI framework organized around protecting Marylanders, centering workers, and keeping children safe. It combines planned legislative recommendations with forthcoming agency guidance and administrative action, while also summarizing existing Maryland requirements.
News summaries are informational and not legal advice. Regulatory status changes frequently.
