One law covers the EU; in the US, the states regulate piece by piece
The AI Act is horizontal legislation that regulates all AI systems by risk tier. At federal level the US uses executive orders to promote innovation and a single framework, while the duties companies must actually meet are spread across state laws and individual federal statutes, and are still shifting through litigation.
Federal: executive orders set direction but do not impose duties directly
Three executive orders
EO 14179 (23 January 2025) removed some of the previous administration’s AI policy barriers; EO 14365 (11 December 2025) calls for a single federal AI policy framework and set up a Justice Department AI litigation task force to challenge conflicting state laws; EO 14409 (2 June 2026) focuses on advanced AI and the cybersecurity of federal systems.
An executive order does not void state law
A state law is displaced only by federal legislation or a court ruling. In April 2026 the Justice Department intervened for the first time, in xAI’s suit against Colorado’s earlier AI law, and a federal court paused enforcement of that law on 27 April 2026; federal preemption bills in Congress have not passed.
The duties that actually apply, by type
Frontier model safety and transparency
California SB 53 took effect on 1 January 2026: frontier model developers must publish a safety framework and transparency reports, report critical safety incidents and protect whistleblowers. The EU counterpart is the AI Act’s duties for general-purpose AI models with systemic risk, with different thresholds and audiences.
Training data disclosure
California AB 2013 took effect on 1 January 2026 and requires generative AI developers to post a summary of their training datasets. xAI’s request to pause enforcement was denied on 4 March 2026 and is on appeal.
Labelling synthetic content
California SB 942 took effect on 2 August 2026 and requires large generative AI services to provide watermarking and detection tools; the transparency duties in Article 50 of the EU AI Act apply from the same day.
Prohibited uses and notice
Texas TRAIGA took effect on 1 January 2026; it prohibits developing or deploying AI with certain harmful intent and requires notice to users by government agencies and in healthcare.
Automated decisions
Colorado SB 26-189, signed on 14 May 2026 and effective 1 January 2027, replaces an earlier law that never took effect: using automated decision-making technology in consequential decisions such as employment, housing, healthcare and insurance requires advance notice, an explanation afterwards and an appeal. The EU counterparts are the AI Act’s Annex III high-risk systems and Article 22 of the GDPR.
Intimate images and deepfakes
The TAKE IT DOWN Act has penalised publishing non-consensual intimate images since 19 May 2025; since 19 May 2026 platforms must remove them within 48 hours of a valid notice, enforced by the Federal Trade Commission.
EU timetable after the AI Omnibus
Already applying
2 February 2025: prohibited practices and AI literacy; 2 August 2025: general-purpose AI model duties; 2 August 2026: transparency duties and most other provisions.
Postponed
The AI Omnibus, Regulation (EU) 2026/1744, took effect on 27 July 2026: Annex III high-risk systems move to 2 December 2027, and high-risk systems embedded in Annex I products to 2 August 2028.
One governance system for both sides
ISO/IEC 42001 and the NIST AI RMF
Build the AI management system on ISO/IEC 42001, define risk with the NIST AI RMF (AI 100-1) and the Generative AI Profile (AI 600-1), and add ISO/IEC 23894 risk management and ISO/IEC 42005 impact assessment.
A certificate is not compliance
A certificate shows that a management system exists; it does not show compliance with the AI Act or any state law. Advanced advisory maps the same evidence to each law’s specific duties.
Evidence the advanced service delivers
- AI system register: purpose, role (developer or deployer), applicable state laws and EU risk tier
- Model and training data descriptions, including sources and licences
- User notices and synthetic content labelling
- Advance notice, explanation and appeal process for automated decisions
- Reporting and removal procedures for critical safety incidents and unlawful content
- Records tracking regulatory change and litigation
Frequently asked questions
Does the US have a federal law like the AI Act?
No. As of 18 November 2026 there is no horizontal federal AI market-access law; executive orders set the direction, and binding duties come from state laws and individual federal statutes such as the TAKE IT DOWN Act.
Does EO 14365 void state AI laws?
Not automatically. The order directs the Justice Department to challenge conflicting state laws and calls for a federal legislative proposal; whether a state law falls depends on Congress legislating preemption or on court rulings.
If we hold ISO/IEC 42001, do we comply with the AI Act and state laws?
No. ISO/IEC 42001 shows that an AI management system exists and works; it is foundation-level evidence. The AI Act and state laws impose specific duties, such as technical documentation, notices and reporting procedures, which must be checked and filled article by article.
Other dimensions of the US trusted supply chain
Regulatory status updated 2026-11-18 · The table maps governance aims and control measures; it does not mean EU and US laws can replace each other legally. Status as of 18 November 2026. Not legal advice.
Confirm which laws actually apply to your products, customers and markets before deciding which evidence to add.
Book an applicability review