IEEE Ethically Aligned Design: 8 Principles and the 7000-Series

IEEE Ethically Aligned Design (EAD), First Edition (2019), sets out eight general principles for building autonomous and intelligent systems, and the IEEE 7000-series turns those principles into testable engineering standards your teams can actually certify against. Where most AI ethics guidance stops at values, IEEE gives engineers a process to follow and a bar to clear.

What is IEEE Ethically Aligned Design?

Ethically Aligned Design is the body of work produced by the IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems. The First Edition was published in 2019 after several earlier draft releases and a long public-consultation period that drew on contributors from engineering, philosophy, law, and the social sciences (OCEANIS announcement).

The full title tells you the intent: "Ethically Aligned Design: A Vision for Prioritizing Human Well-being with Autonomous and Intelligent Systems." It reads as a treatise, not a checklist. The practical teeth come from the partner work, the IEEE 7000-series of standards, which exist to make the principles measurable.

That split matters. EAD is the philosophy. The 7000-series is how you build to it.

What are the eight principles of IEEE Ethically Aligned Design?

EAD is organised around eight General Principles for the design, development, deployment, and decommissioning of autonomous and intelligent systems (A/IS). The list below mirrors the First Edition document (EAD1e, IEEE).

Principle	What it asks of builders
Human Rights	A/IS are created and operated to respect, promote, and protect internationally recognised human rights.
Well-being	Creators adopt increased human well-being as a primary success criterion.
Data Agency	People can access and securely share their data, keeping control over their identity.
Effectiveness	Creators provide evidence that a system works as intended and meets its purpose.
Transparency	The basis of a decision is discoverable.
Accountability	Creators and operators are answerable for system decisions and consequences.
Awareness of Misuse	Creators guard against the risks of misuse once a system is in the world.
Competence	Operators have the knowledge and skill to run a system safely.

A few of these break from the usual ethics-framework wording in a way I think is the point. "Well-being" as a primary success criterion, not just safety, sets a higher bar than "do no harm." And "Competence" puts a duty on operators, not only designers, which most frameworks skip.

What are the IEEE 7000-series standards?

This is where EAD stops being a vision and starts being engineering work. Each 7000-series standard takes an ethical concern and defines a process, with requirements you can audit. Here are the published ones most boards and regulated teams ask about.

Standard	Title	Status
IEEE 7000-2021	Model Process for Addressing Ethical Concerns During System Design	Published 2021
IEEE 7001-2021	Transparency of Autonomous Systems	Published 2021
IEEE 7002-2022	Data Privacy Process	Published 2022
IEEE 7003-2024	Algorithmic Bias Considerations	Published (released January 2025)
IEEE 7007-2021	Ontological Standard for Ethically Driven Robotics and Automation Systems	Published 2021
IEEE 7010-2020	Recommended Practice for Assessing the Impact of Autonomous and Intelligent Systems on Human Well-Being	Published 2020

IEEE 7000-2021 is the anchor. It was approved in June 2021 and published in September 2021, and it sets out the process organisations follow to surface ethical values early in design and trace them through to operational concepts (IEEE SA). One hundred and fifty-four experts took part in its development, with the balloting group voting it through (IEEE Technology and Society). It later fed into the joint ISO/IEC/IEEE 24748-7000:2022 standard.

IEEE 7001-2021 is worth singling out. It defines measurable, testable levels of transparency so an autonomous system can be assessed objectively and a compliance level determined (IEEE Xplore). Transparency you can measure, not transparency you assert.

How does IEEE 7000 differ from NIST, ISO, and OECD?

These frameworks aren't rivals. They sit at different altitudes, and the smart move is to stack them.

• NIST AI RMF gives you the risk-management lifecycle. IEEE 7001 and 7003 give you the technical depth to satisfy the transparency and bias functions inside it.
• ISO/IEC 42001 is the management-system layer, the governance scaffolding around AI. IEEE standards can serve as the concrete engineering mechanisms that an ISO management system points to.
• OECD AI Principles are high-level and government-facing. IEEE gives engineers a buildable route to honour them.

Put plainly: OECD says what good looks like, NIST and ISO say how to manage and govern it, and IEEE 7000-series says how to build and test it. A regulated business deploying AI usually needs all three altitudes covered.

How do you put IEEE EAD into practice?

You don't adopt all of EAD at once. You pick the standards that match your risk and work outward.

1. Map your highest-risk AI use cases to the relevant standards. A credit-decisioning model points you at IEEE 7003 (bias). A customer-facing autonomous system points you at IEEE 7001 (transparency) and IEEE 7002 (data privacy).
2. Run IEEE 7000-2021 as the wrapper process. It's the standard that builds ethics elicitation into the design lifecycle rather than bolting it on at the end.
3. Define the evidence up front. Each standard expects records: design decisions, transparency levels, bias controls. Decide who owns that documentation before you start, not after an auditor asks.
4. Build review gates at design milestones, so ethical requirements get checked at the same points you check functional ones.
5. Re-baseline as the standards update. IEEE 7003 only landed for general use in early 2025. The series is still maturing, so treat your selection as a living list.

The honest difficulty here is documentation burden and the expertise gap. Most teams have engineers who understand the systems and a governance function that understands the ethics, and the two rarely speak the same language. Closing that gap is the real work. The standards just give both sides a shared vocabulary.

Frequently asked questions

Is IEEE Ethically Aligned Design a standard you can certify against?

EAD itself is a guidance document, not a certifiable standard. The certifiable parts are the IEEE 7000-series standards that operationalise it, such as IEEE 7000-2021 and IEEE 7001-2021. IEEE also runs a separate conformity-assessment programme, CertifAIEd, built on this body of work.

When was IEEE Ethically Aligned Design published?

The First Edition was published in 2019, following earlier public-comment drafts released from 2016 onwards (NSPE). The 7000-series standards have been published on a rolling basis since 2020.

Which IEEE 7000-series standard covers AI bias?

IEEE 7003-2024, Algorithmic Bias Considerations. It sets out processes to define, measure, and mitigate algorithmic bias, including criteria for selecting validation datasets and guidance on communicating an algorithm's intended application boundaries (IEEE Xplore). It was released for general use in January 2025.

Does IEEE 7000 replace NIST or ISO?

No. IEEE 7000-series standards complement them. NIST AI RMF handles risk management, ISO/IEC 42001 handles the management system, and IEEE provides the technical engineering depth that sits underneath both.

The bottom line

IEEE's contribution is the one I'd push regulated clients towards first when their engineers ask "but what do we actually build?" EAD sets the values, and the 7000-series gives those values a process and a measurable bar, which is more than most AI ethics guidance manages. The catch is that the series is still filling out, with a key bias standard only general-use since early 2025, so anyone treating it as a finished checklist will get caught out. Pick the standards that match your risk, build the documentation discipline early, and stack IEEE underneath your NIST and ISO work rather than choosing between them.

This is the kind of work our responsible AI governance handles.