NORBERT BIEDRZYCKI

The First Global AI Ethics Framework: Can the World Agree on Responsible AI?

The First Global AI Ethics Framework: Can the World Agree on Responsible AI?

Last month, the 193 member states of UNESCO adopted the Recommendation on the Ethics of Artificial Intelligence. It is the first global normative framework on AI ethics agreed by governments rather than by companies or experts. It covers fairness, transparency, privacy, human oversight, environmental sustainability and the protection of human dignity, and it grounds all of them in human rights.

It is a genuine milestone. It is also worth being clear about what it is and what it is not.

What changed

Until now the AI ethics debate was led by technology companies, universities and civil society organizations. Governments participated, but mostly as commentators. The UNESCO recommendation signals that AI ethics has become a matter of international public policy.

That matters for a practical reason. Principles adopted by governments eventually find their way into procurement rules, regulatory guidance and national strategies. Companies selling AI to the public sector will feel this first.

What did not change

A recommendation is not a law. It has no enforcement mechanism. And agreement on principles at a global level is possible precisely because principles are general enough for everyone to interpret them differently.

Consider privacy. The European Union treats it as a fundamental right, enforced through GDPR. The United States approaches it largely through sector rules and consumer protection. China has just introduced its own Personal Information Protection Law, while maintaining a very different relationship between the state and personal data. All three can sign the same document about privacy. They do not mean the same thing.

The same applies to human oversight, transparency and fairness. The words travel. The interpretations do not.

The fragmentation risk for business

For multinational companies, this creates a specific challenge. There will not be one global standard for responsible AI. There will be several overlapping regimes built on similar vocabulary and different legal traditions.

The companies that navigate this best will treat global principles as a floor and design their internal governance to meet the strictest regime in which they operate. That is more expensive in the short term and much cheaper than rebuilding systems market by market later. It is exactly the logic many companies applied to GDPR after 2018.

There is also a deeper point. Ethical frameworks reflect values, and values reflect political systems. A global consensus on AI ethics will remain partial for as long as societies disagree about freedom, privacy and the role of the state. Companies operating across those systems will increasingly have to decide which values they embed in their products, rather than assuming the question has been settled for them.

Leadership question: Can responsible AI become a universal standard when governments and companies have fundamentally different views on privacy, freedom and technological development?

Read next

  1. Ethical AI: From Principles to Accountability, 2020–2026

  2. The Ethics of AI and Human Autonomy: Are We Giving Away Too Much Control?

    Ethical AI series, written in 2026

  3. The Accountability Gap: Who Is Responsible When AI Makes Decisions Independently?

    Ethical AI series, written in 2026