Where the rules stand right now.
Plain-English status on the regulations our clients ask about most, with key dates, what to do next, and links to the official sources.
U.S. federal AI policy: America’s AI Action Plan and executive orders
The United States has no single federal artificial intelligence (AI) statute. Federal policy runs through executive orders (EOs), the White House’s America’s AI Action Plan, Office of Management and Budget (OMB) rules for federal agencies, and existing laws that regulators such as the Federal Trade Commission (FTC) already enforce. The current direction favors innovation and one national standard over a patchwork of state rules, but consumer-protection, civil-rights, health, and financial rules still apply to AI.
- JAN 23, 2025EO 14179 revokes the 2023 AI executive order (EO 14110) and sets a pro-innovation direction
- APR 3, 2025OMB memos M-25-21 and M-25-22 set rules for how federal agencies use and buy AI, which flow down to vendors through contracts
- JUL 23, 2025America’s AI Action Plan: 90+ federal actions under three pillars (accelerate AI innovation, build American AI infrastructure, lead in international AI diplomacy and security), plus companion orders on AI exports, data-center permitting, and ideological neutrality for AI models the government buys
- DEC 11, 2025EO 14365, “Ensuring a National Policy Framework for AI”: a Department of Justice (DOJ) AI Litigation Task Force (stood up January 2026) to challenge state AI laws, and a Commerce review of “onerous” state laws
- MAR 20, 2026White House National Policy Framework for AI: nonbinding recommendations to Congress, including targeted preemption of state AI laws
- MAY 19, 2026TAKE IT DOWN Act platform duties begin: remove nonconsensual intimate images, including AI deepfakes, within 48 hours of a valid request (FTC enforces)
- JUN 2, 2026EO 14409, “Promoting Advanced AI Innovation and Security”: AI-enabled cyber defense, a Treasury-led AI cybersecurity clearinghouse, and voluntary pre-release government testing of frontier models
- SEP 29, 2026EO 14434: federal agencies now say “Super Intelligence (SI)” instead of “AI” in their communications. Terminology only; statutory definitions and private-sector duties are unchanged
What it means for your business
- No federal AI license or certification exists, but the FTC Act still covers deceptive AI marketing claims (“AI washing”) and unfair data practices
- Selling to government? OMB M-25-21 and M-25-22 shape AI terms in federal contracts: expect documentation, testing, and data-use requirements
- Anti-discrimination, lending, employment, and health care rules apply to AI-driven decisions the same way they apply to human ones
- State AI laws stay enforceable unless Congress preempts them or a court blocks them, so plan for both
- Watch next: a proposed federal “Super Intelligence” definition (due late November 2026) and any preemption bill in Congress
Our view: preemption fights won’t stop the FTC, state attorneys general, or plaintiffs. A documented AI governance program aligned to the NIST AI RMF is the defensible baseline either way.
NIST AI Risk Management Framework (AI RMF 1.0)
The National Institute of Standards and Technology (NIST) AI RMF is the voluntary framework U.S. regulators, buyers, and courts look to for “reasonable” AI risk management. Texas’s AI law, for example, gives weight to following it. It is organized into four functions. Click one:
- JAN 26, 2023AI RMF 1.0 and companion Playbook released
- JUL 26, 2024Generative AI Profile (NIST AI 600-1): 12 risks unique to or worsened by generative AI, with suggested actions
- MAR 2025Adversarial machine learning taxonomy (NIST AI 100-2 E2025): standard terms for attacks such as poisoning, evasion, and prompt injection
- JUL 2025AI Action Plan directs NIST to revise the AI RMF
- DEC 2025Cyber AI Profile (NIST IR 8596) preliminary draft maps AI onto Cybersecurity Framework (CSF) 2.0: securing AI systems, using AI for defense, and thwarting AI-enabled attacks
- 2026AI RMF revision underway; SP 800-53 control overlays for securing AI systems (COSAiS) in development; critical infrastructure profile concept note (April 7, 2026)
How to put it to work
- Inventory every AI system and use, including “shadow AI” and AI features built into the SaaS tools you already pay for
- Assign an owner for each system and publish an AI acceptable-use policy (Govern)
- Rate each use case by impact and map the data and people it touches (Map)
- Test for accuracy, bias, prompt injection, and data leakage before launch and after changes (Measure)
- Track incidents, vendor model updates, and retirement of systems (Manage)
- Pair it with CSF 2.0 for security controls, and ISO/IEC 42001 if customers want a certifiable AI management system
Smart move: build one program on the AI RMF and map it to the EU AI Act and state laws, instead of chasing each rule separately.
California
California has more AI-specific rules than any other state. Most apply based on where your customers, users, or employees are, not where your company is based.
- OCT 1, 2025Civil Rights Council rules: automated decision systems used in hiring and employment fall under anti-discrimination law, with records kept for four years
- JAN 1, 2026SB 53 Transparency in Frontier AI Act (large frontier developers publish a safety framework and report critical safety incidents); AB 2013 training-data disclosure for generative AI developers; SB 243 companion-chatbot safeguards; and California Consumer Privacy Act (CCPA) regulations on automated decision-making technology (ADMT), risk assessments, and cybersecurity audits take effect
- AUG 2, 2026SB 942 AI Transparency Act (as amended by AB 853): generative AI providers with 1 million+ monthly users must offer a free AI-detection tool and embed disclosures in AI-generated content
- SEP 30, 2026SB 947 “No Robo Bosses Act” signed: a human must review AI-driven discipline or termination decisions (operative July 1, 2027)
- JAN 1, 2027CCPA ADMT compliance date: pre-use notice, opt-out, and access rights when ADMT makes significant decisions about consumers
Other states to watch
- ColoradoSB 26-189 (signed May 14, 2026) repealed and replaced the original 2024 Colorado AI Act after a federal court paused its enforcement. Starting January 1, 2027: notice, explanation, data-correction, and human-review rights when automated decisions affect jobs, lending, housing, insurance, health care, or education. Attorney General enforcement with a 60-day cure period. A separate Chatbot Safety Act (HB 26-1263) adds AI disclosure and teen protections for chatbot operators.SB 26-189 ↗Colorado AG: AI rules & rulemaking ↗
- TexasTexas Responsible AI Governance Act (TRAIGA, HB 149), in effect since January 1, 2026: bans certain harmful AI uses and requires AI disclosures by government agencies and health care providers. Attorney General enforcement with a 60-day cure period; following the NIST AI RMF supports a defense.HB 149 ↗Texas AG: AI rights & complaints ↗
- IllinoisHB 3773, in effect since January 1, 2026: AI that produces discrimination in hiring, promotion, or discipline is a civil-rights violation, and employees must be told when AI is used.Public Act 103-0804 (HB 3773) ↗
- New YorkRAISE Act (amended March 2026), effective January 1, 2027: safety frameworks and incident reporting for large frontier AI developers. New York City Local Law 144 already requires bias audits and notices for automated hiring tools.RAISE Act (S6953) ↗NYC Local Law 144 ↗
- UtahAI Policy Act (2024): using generative AI is no defense to a consumer-protection violation; disclose AI use when asked, and up front in licensed professions.SB 149 (2024) ↗
What to do now
- Map where your customers, applicants, and employees live: state AI laws follow the people, not your headquarters
- Inventory AI that informs decisions about jobs, lending, housing, insurance, health care, or education; that is where most state rules land
- Add AI-interaction notices, plain-language explanations for adverse decisions, and a path to human review
- Get AI vendors to commit in contract to documentation, bias testing, incident notice, and limits on data use
- Anchor everything in one NIST AI RMF-based program so a single set of evidence answers many states at once
Federal preemption is being pushed but has not been enacted. As of November 2026, state AI laws remain enforceable unless a court blocks them, as happened with Colorado’s original law in April 2026.
EU Cyber Resilience Act — Regulation (EU) 2024/2847
The Cyber Resilience Act (CRA) sets mandatory cybersecurity requirements for “products with digital elements” (hardware and software) sold in the European Union (EU), across their whole lifecycle. It applies to manufacturers, importers, and distributors, including U.S. companies selling into the EU. Fines reach up to €15 million or 2.5% of worldwide annual turnover.
- DEC 10, 2024Entered into force
- JUN 11, 2026Rules for conformity assessment (notified) bodies apply
- SEP 11, 2026Manufacturers must report actively exploited vulnerabilities and severe incidents: 24-hour early warning, 72-hour notification, then a final report (14 days after a fix for vulnerabilities; one month for incidents) via the Single Reporting Platform run by the European Union Agency for Cybersecurity (ENISA)
- DEC 11, 2027Full application: essential requirements, conformity assessment, technical documentation, and CE marking
What to do now
- Inventory every product with digital elements you place on the EU market
- Stand up a 24/72-hour vulnerability and incident reporting runbook
- Publish a coordinated vulnerability disclosure policy and generate software bills of materials (SBOMs)
- Map secure-by-design and update obligations into your product lifecycle
- Plan conformity assessment and CE marking for the December 2027 deadline
CMMC 2.0: where things stand
On July 13, 2026, the Department of War (DoW, formerly the Department of Defense) paused Phase 2, which would have made Certified Third-Party Assessment Organization (C3PAO) Level 2 certification the default for many contracts starting November 10, 2026. It also launched a 60-day Cybersecurity Maturity Model Certification (CMMC) Reform Task Force and a public request for information. As of late October 2026, no resumption date has been announced. The 32 Code of Federal Regulations (CFR) Part 170 rule itself remains on the books.
- DEC 16, 2024CMMC program rule (32 CFR Part 170) takes effect
- NOV 10, 2025Phase 1 begins: Level 1 and Level 2 self-assessments in contracts (Defense Federal Acquisition Regulation Supplement (DFARS) rule)
- JUL 13, 2026Phase 2 (C3PAO by default) suspended; reform task force and request for information (RFI) launched
- SEP 2026Class deviation tells contracting officers to accept self-assessments and remove third-party assessment requirements at the next modification
- OCT 2026Task force review closed; final determinations pending
What still applies today
- DFARS 252.204-7012 safeguarding and 72-hour incident reporting
- National Institute of Standards and Technology (NIST) Special Publication (SP) 800-171 Rev. 2 (110 requirements) for systems handling controlled unclassified information (CUI)
- Level 1 / Level 2 self-assessments, Supplier Performance Risk System (SPRS) scores, and annual senior-official affirmations
- Existing C3PAO certifications remain valid for their three-year term
- False Claims Act exposure for inaccurate affirmations continues
Our view: the pause is a window to get audit-ready, not a reason to stop. Requirements can return quickly.
HIPAA Security Rule
The current Health Insurance Portability and Accountability Act (HIPAA) Security Rule remains in force. U.S. Department of Health and Human Services (HHS) proposed the first major update in two decades (published January 6, 2025). The federal regulatory agenda now lists final action no earlier than mid-2027, and agenda dates are not binding.
- JAN 6, 2025Proposed rule published in the Federal Register
- MAR 7, 2025Public comment period closed
- 2026Final rule delayed; current rule still enforced by HHS Office for Civil Rights (OCR)
What the proposal would require
- Remove “addressable” flexibility: multi-factor authentication (MFA), encryption, and segmentation become expected
- Written asset inventory and network map for systems touching electronic protected health information (ePHI)
- Risk analysis reviewed at least every 12 months
- Vulnerability scans every 6 months and penetration tests every 12 months
- Ability to restore critical systems within 72 hours
Smart move: adopt HHS 405(d) Health Industry Cybersecurity Practices now; they line up with where the rule is heading.
PCI DSS v4.0.1
Version 4.0.1 is the active standard for anyone who stores, processes, or transmits cardholder data. The “future-dated” requirements introduced in v4.0 became mandatory on March 31, 2025, so assessments now test all of them.
- MAR 31, 2024PCI DSS v3.2.1 retired
- JUN 2024v4.0.1 published (clarifications, no new requirements)
- MAR 31, 2025Future-dated requirements became mandatory
Where teams most often fall short
- MFA for all access into the cardholder data environment
- Targeted risk analyses to justify control frequencies
- Inventory and integrity monitoring of payment-page scripts (e-commerce skimming)
- Authenticated internal vulnerability scanning
- Stronger password and service-account management
NIST Risk Management Framework (SP 800-37 Rev. 2)
The Risk Management Framework (RMF) is the seven-step process federal agencies and many contractors use to manage security and privacy risk across a system's lifecycle. Click a step:
Direct links
- RMF overview, quick-start guides, and FAQs
- SP 800-37 Rev. 2: the RMF process itself
- SP 800-53 Rev. 5: the control catalog used in the Select step
- Pairs with NIST artificial intelligence (AI) RMF for AI systems, and with Cybersecurity Framework (CSF) 2.0 for enterprise programs
Last reviewed November 3, 2026. This summary is general information, not legal advice; confirm requirements against the official sources.
