The Blueprint for an AI Bill of Rights: What It Says

The Blueprint for an AI Bill of Rights is a 2022 White House white paper that set out five principles for automated systems. It is not a law, it was not enforceable when it was written, and it is not any administration’s current federal AI policy today. This guide covers what it actually says, what “non-binding” means in practice, and where its influence shows up in the AI policy conversation that followed it — including in the binding state law covered elsewhere in this cluster, California’s SB 53.

The short answer

The White House Office of Science and Technology Policy (OSTP) released the Blueprint for an AI Bill of Rights: Making Automated Systems Work for the American People on October 4, 2022. It is a white paper, not legislation and not an executive order. It sets out five principles meant to guide the design, use, and deployment of automated systems, plus a companion “From Principles to Practice” handbook of implementation suggestions. The document says explicitly, in its own “About This Document” section, that it “is non-binding and does not constitute U.S. government policy” and that it “does not supersede, modify, or direct an interpretation of any existing statute, regulation, policy, or international instrument.” No agency or company was ever required to comply with it.

The five principles

The Blueprint states five principles, each given here in the document’s own words:

1. Safe and Effective Systems. “You should be protected from unsafe or ineffective systems.”
2. Algorithmic Discrimination Protections. “You should not face discrimination by algorithms and systems should be used and designed in an equitable way.”
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.”
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.”
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.”

What the principles ask for in practice

The companion “From Principles to Practice” section attaches more specific expectations to each principle, though none of them are requirements in the legal sense. For Algorithmic Discrimination Protections, it calls for proactive equity assessments during design, use of “representative and robust data” with testing for demographic proxies, ongoing disparity assessment and mitigation after deployment, and independent evaluation and public reporting through algorithmic impact assessments. For Data Privacy, it calls for privacy-by-design and data minimization, use-specific consent requested in “plain language,” individual access to one’s own data, and heightened limits on surveillance in sensitive contexts such as workplaces and schools. The other three principles are stated at a higher level in the source document itself: protection from unsafe or ineffective systems, transparency about when an automated system is in use and why, and a human fallback path when something goes wrong.

Why “non-binding” matters here

This is the point that most distinguishes the Blueprint from the state laws covered elsewhere in this cluster. It created no cause of action, no regulator, and no penalty. It was persuasive guidance aimed at agencies, companies, and the public conversation, not a statute a company could violate. That is a direct contrast with laws such as California’s SB 53, which does impose statutory disclosure, incident-reporting, and enforcement requirements on covered developers. The gap between the Blueprint’s five aspirational principles in 2022 and binding statutory text a few years later is itself part of the story of how U.S. AI policy has developed.

Its status now

The Blueprint was a Biden administration OSTP publication; it was never re-issued or adopted by statute, so it doesn’t carry over automatically from one administration to the next the way a law would. Its practical influence peaked in October 2023, when President Biden’s Executive Order 14110 on AI stated the administration would “build on the important steps that have already been taken — such as issuing the Blueprint for an AI Bill of Rights, the AI Risk Management Framework, and Executive Order 14091.” That EO was itself revoked on January 23, 2025, by a Trump administration order, “Removing Barriers to American Leadership in Artificial Intelligence,” as part of a broader shift away from the prior administration’s AI policy. The Blueprint’s own page no longer exists on the current whitehouse.gov site (it now returns a 404) and it is not listed among the current administration’s AI policy documents on ai.gov, which instead point to America’s AI Action Plan and a newer set of executive orders and OMB memoranda. In practice, the Blueprint is a historical document today: it is preserved at the Biden administration’s archived OSTP site, and it is not current federal AI policy.

How CASRAI’s NIKOLAI tracks this

The Blueprint’s “Human Alternatives, Consideration, and Fallback” principle — that people affected by an automated system should be able to reach a person who can quickly consider and remedy problems — describes the same governance shape CASRAI catalogs independently as the Accountable Decision-Maker and Sign-Off element in NIKOLAI’s Commitments and Governance track (N9). NIKOLAI documents how individual frontier AI developers name a specific role or person responsible for high-stakes safety and deployment decisions, years after the Blueprint first gestured at that same idea for automated systems generally. NIKOLAI is CASRAI’s own, unendorsed reference project — it does not implement the Blueprint or represent any developer’s or agency’s official position — but it is one place CASRAI tracks how that accountability principle shows up in practice. See NIKOLAI’s Commitments and Governance track.

FAQ

Is the Blueprint for an AI Bill of Rights a law?

No. It is a White House OSTP white paper published in October 2022. It was never introduced as legislation and never issued as an executive order, and its own text says it “does not constitute U.S. government policy” and creates no compliance obligation.

Is it still current federal policy?

No. It is no longer hosted on the current whitehouse.gov site and is not referenced among the current administration’s AI policy documents, which center on America’s AI Action Plan and later executive orders and OMB guidance. The 2023 executive order that had explicitly built on it was revoked in January 2025.

Did any binding law come directly from the Blueprint?

Not as a direct legal source. The Blueprint is a set of principles, not model statutory text. What can be verified is that it was cited as a building block in Executive Order 14110 (2023), alongside the NIST AI Risk Management Framework; separately, states went on to pass their own binding AI statutes covering some of the same ground, such as algorithmic discrimination and transparency, on their own legislative track.

What is the difference between the five principles and the “From Principles to Practice” section?

The five principles are the short, plain-language statements quoted above. “From Principles to Practice” is a longer companion section of the same document that attaches more specific, still non-binding, implementation suggestions to each principle.

Sources

• White House OSTP, Blueprint for an AI Bill of Rights: Making Automated Systems Work for the American People, October 4, 2022 (archived).
• White House OSTP, “About This Document,” Blueprint for an AI Bill of Rights (archived).
• Executive Order 14110, “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence,” October 30, 2023 (archived).
• Executive Order, “Removing Barriers to American Leadership in Artificial Intelligence,” January 23, 2025.