The Blueprint for an AI Bill of Rights in 2026: What It Said and Which Laws Now Require Something Similar

Is the AI Bill of Rights a law?

No. The Blueprint for an AI Bill of Rights never created a legal duty for any company or agency. OSTP published it on October 4, 2022 as “Blueprint for an AI Bill of Rights: Making Automated Systems Work for the American People,” a 73-page document that its own front matter calls “a white paper.” It was issued as a white paper, not as a statute, an executive order, or a regulation.

The legal disclaimer on page 2 leaves no room for a different reading. The Blueprint “is non-binding and does not constitute U.S. government policy.” It does not supersede or modify any statute or regulation, and it “does not constitute binding guidance for the public or Federal agencies and therefore does not require compliance with the principles described herein.” It also states that it does not create any legal right enforceable by any party against the United States or any other person.

No regulator could fine a company for ignoring the Blueprint, and no individual could sue under it. Its influence came through later documents that bound federal agencies. State, city, and EU laws that impose similar requirements rest on their own legal authority, not on the Blueprint.

Is the Blueprint still in effect in 2026?

Not as federal policy. As of October 26, 2026, we found no current executive order or OMB memorandum that relies on it, and the order and memorandum that did were revoked and rescinded in 2025.

Whether it was formally withdrawn is a separate question, and the primary record does not show a withdrawal. We found no executive order, OMB memorandum, or OSTP notice that rescinds the Blueprint by name. Neither Executive Order 14148, which revokes a list of orders and other actions by name, nor Executive Order 14179 mentions the Blueprint. What the record shows instead:

• The address printed in the Blueprint itself, whitehouse.gov/ostp/ai-bill-of-rights, returned an HTTP 404 (not found) error when we checked on October 26, 2026.
• The full Blueprint and its technical companion remain on the archived White House website for the 2021 to 2025 administration, which carries the banner “This is historical material ‘frozen in time’.”
• Neither M-25-21 nor M-25-22, the 2025 memoranda that now govern federal agency use and purchase of AI, mentions the Blueprint.

What are the five principles of the AI Bill of Rights?

The Blueprint applies a two-part test: it covers automated systems that have the potential to meaningfully impact the American public’s rights, opportunities, or access to critical resources or services. Each principle opens with a one-sentence statement written to the person affected:

1. Safe and Effective Systems. “You should be protected from unsafe or ineffective systems.” It asks for consultation with diverse communities, stakeholders, and domain experts, pre-deployment testing, risk mitigation, ongoing monitoring, and the option of not deploying a system at all.
2. Algorithmic Discrimination Protections. “You should not face discrimination by algorithms and systems should be used and designed in an equitable way.” It asks for equity assessments, representative data, disparity testing before and after deployment, and a public algorithmic impact assessment whenever possible.
3. Data Privacy. “You should be protected from abusive data practices via built-in protections and you should have agency over how data about you is used.” The expectations are privacy by default, collection limited to what is strictly necessary, plain-language consent, and heightened protections in sensitive domains such as health, work, and finance.
4. Notice and Explanation. “You should know that an automated system is being used and understand how and why it contributes to outcomes that impact you.” In practice that means plain-language documentation, notice that a system is in use, and explanations of outcomes that are clear, timely, and accessible.
5. Human Alternatives, Consideration, and Fallback. “You should be able to opt out, where appropriate, and have access to a person who can quickly consider and remedy problems you encounter.” It asks for a human alternative where appropriate and a fallback process when a system fails or a person wants to appeal.

Most of the document, pages 12 to 73, “From Principles to Practice: A Technical Companion to the Blueprint for an AI Bill of Rights,” explains for each principle why it matters, what should be expected of automated systems, and how the principles can move into practice. It reads as a design checklist, not a legal standard.

How did the Blueprint shape federal AI policy in 2023 and 2024?

It worked through an executive order and an OMB memorandum that directed federal agencies.

Executive Order 14110, signed October 30, 2023, named the Blueprint in Section 2(d), its statement of policy on equity and civil rights, as one of the steps already taken that the administration would build on. Section 10.1(b) then directed OMB to issue guidance requiring “minimum risk-management practices for Government uses of AI that impact people’s rights or safety,” including, where appropriate, “practices derived from OSTP’s Blueprint for an AI Bill of Rights and the NIST AI Risk Management Framework.” The order listed them: public consultation, data quality assessment, assessing and mitigating disparate impacts and algorithmic discrimination, notice of the use of AI, continuous monitoring, and human consideration and remedies for adverse decisions.

OMB issued that guidance as M-24-10 on March 28, 2024. It required agencies, by December 1, 2024, to apply minimum practices to “safety-impacting” and “rights-impacting” AI or stop using it. For rights-impacting AI, the extra practices included assessing impacts on equity and fairness, consulting affected communities, notifying negatively affected individuals, maintaining human consideration and remedy processes, and maintaining options to opt out of AI-enabled decisions where practicable. It also named the Blueprint as a source of additional best practices. M-24-18, issued September 24, 2024, set rules for how agencies buy AI.

Both memoranda governed federal agencies’ own use and purchase of AI. Neither placed duties directly on private companies; a company met them mainly as a vendor to an agency, or as the subject of an agency decision made with AI.

What changed in 2025 and 2026?

Five documents reset the 2023 and 2024 framework.

• Executive Order 14148 (January 20, 2025) revoked Executive Order 14110 by name, in Section 2(ggg).
• Executive Order 14179 (January 23, 2025), “Removing Barriers to American Leadership in Artificial Intelligence,” directed a review of all actions taken under the revoked order and told the OMB Director to revise M-24-10 and M-24-18 within 60 days, as necessary, to make them consistent with its policy.
• OMB Memoranda M-25-21 and M-25-22 (April 3, 2025). M-25-21 states that it “rescinds and replaces” M-24-10, and M-25-22 states that it rescinds and replaces M-24-18. Both remained listed on OMB’s memoranda page as of October 26, 2026.
• America’s AI Action Plan (July 23, 2025) listed rescinding Executive Order 14110 among steps already taken and recommended that OMB work with federal agencies to “identify, revise, or repeal” regulations, guidance, and other documents “that unnecessarily hinder AI development or deployment.” It does not mention the Blueprint.

M-25-21 replaced the two categories of rights-impacting and safety-impacting AI with one: “high-impact AI,” meaning AI whose output serves as a principal basis for decisions or actions with legal, material, binding, or significant effect on matters such as civil rights, civil liberties, or privacy; access to education, housing, insurance, credit, employment, or critical government services; human health and safety; or critical infrastructure and public safety. For that category, agencies must apply seven minimum practices: pre-deployment testing, an AI impact assessment with an independent reviewer, ongoing monitoring for performance and adverse impacts, operator training, human oversight with a fail-safe where practicable, “timely human review and a chance to appeal” for people affected by AI-enabled decisions, and a channel for feedback from end users and the public. Agencies had 365 days from issuance to document implementation, and must stop using a non-compliant high-impact use case.

So testing, impact assessment, monitoring, human oversight, and human review on appeal remain in federal agency rules. Two items on M-24-10’s rights-impacting list do not appear among M-25-21’s seven practices: notifying negatively affected individuals, and offering an opt-out from AI-enabled decisions. M-25-21 mentions opting in or out only for pilot programs.

At least one agency also edited guidance that cited the Blueprint. FDA’s guidance on predetermined change control plans for AI-enabled devices, as issued on December 4, 2024, said it was “consistent with and promotes the principles described in the Blueprint for an AI Bill of Rights.” The version FDA reissued on August 18, 2025, the one on fda.gov today, contains no reference to the Blueprint. Our FDA PCCP guide lists the other changes in that reissue.

Executive Order 14434, signed September 29, 2026, directs executive departments and agencies to use “Super Intelligence” and “SI” in place of “Artificial Intelligence” and “AI” in their own non-statutory documents. It binds those departments and agencies only and does not mention the Blueprint; it changes the vocabulary of later federal documents, not any duty discussed here.

Where do the Blueprint’s principles appear in binding law today?

In state, city, and EU laws that impose similar requirements on their own legal authority. Each reaches a narrower set of decisions than the Blueprint did.

Colorado SB 26-189 applies to consequential decisions made on or after January 1, 2027. A federal court order in X.AI LLC v. Weiser bars the Attorney General from enforcing it, including by investigation, for violations occurring on or before 14 days after the court rules on xAI’s preliminary injunction motion; xAI had not filed that motion as of October 26, 2026. The Colorado SB 26-189 guide covers the duties in detail, and the NYC Local Law 144 guide covers bias audits and candidate notice. California’s rules are in the California AI laws guide, and the EU dates are in the EU AI Act compliance guide.

Did any state pass an “AI Bill of Rights”?

We found no enacted state law with that title as of October 26, 2026. Three bills introduced under that name, one in Oklahoma and two in Florida, each passed one chamber, and the official legislative records show that none became law.

• Oklahoma HB 3453 (2024). As introduced, the bill named itself the Oklahoma Artificial Intelligence Bill of Rights. The House adopted a floor substitute that, according to the House bill summary, removed that name and set out eight provisions, including notice when images or text are generated entirely with AI, protection against algorithmic discrimination, and data privacy. The House passed it 89 to 0 on March 14, 2024. The last action on the Legislature’s bill page is a Senate referral to the Judiciary Committee on March 27, 2024.
• Florida SB 482 and SB 2-D (2026). SB 482, titled “Artificial Intelligence Bill of Rights,” covered government AI contracts, parental consent before a minor holds a companion chatbot account, and periodic notices from bot operators, among other things. The Senate passed it 35 to 2 on March 4, 2026, and it died in House messages on March 13, 2026. A special-session version, SB 2-D, passed the Senate 37 to 1 on April 28, 2026 and died in a House subcommittee on April 29, 2026.

What does this mean for a company in 2026?

The Blueprint itself requires nothing. A company that cites it as a reason for its AI program, or as proof of compliance, is citing a document with no legal force and no current federal standing. The duties that do apply come from laws that reach particular decisions in particular places.

The five principles still work as a map of where those laws concentrate. Most of the entries in the table above fall under the last two principles: notice before use, an explanation after an adverse decision, and a way to reach a human. A practical sequence:

• List each system that makes or materially shapes a decision about a person, and where those people live
• Match each system to the laws that reach it: California now (Civil Rights Council employment rules since October 1, 2025, and CCPA risk assessments since January 1, 2026) and from January 1, 2027 (CCPA automated decisionmaking rules), Colorado from January 1, 2027, NYC for hiring and promotion, the EU AI Act for systems used there, and existing anti-discrimination law everywhere
• Write the pre-use notice and the adverse-decision explanation those laws require, and test them on a real case
• Name who performs human review, what authority they have to change the result, and what records they keep
• If you sell AI to federal agencies, expect agencies working under M-25-21 and M-25-22 to ask for testing, impact assessment, and monitoring information

The AI compliance guide sets out the first 90 days, and the governance frameworks guide explains the NIST AI Risk Management Framework, which Executive Order 14110 named alongside the Blueprint and which remains a current, voluntary NIST framework.

This guide is general information as of October 26, 2026, not legal advice; confirm current requirements with the issuing agency, legislature, or counsel before acting on it.