UN rights council in Geneva backs safeguards for artificial intelligence

The UN Human Rights Council has placed artificial intelligence inside the international human rights framework for the first time. Delegates in Geneva adopted the resolution by consensus on Tuesday, 7 October 2026, warning that AI systems can threaten rights if governments and companies deploy them without adequate controls.

The text applies across the life of an AI system, from its manufacture to its use. It calls for compliance with international law at every stage and says applications that violate human rights should be prohibited.

The decision gives Geneva another role in a rapidly expanding debate over how governments should regulate automated systems. The resolution does not establish a global regulator or set out a binding enforcement mechanism. It does, however, create a common political reference point for states drafting national rules.

For Switzerland, the issue carries particular weight. Swiss institutions, companies and research groups operate in a country that is both a technology hub and a centre for international diplomacy. The Council’s decision places those two roles in direct proximity, with future negotiations expected to return to Geneva.

Human oversight must remain attached to every decision involving AI, the resolution says. Governments are urged to define who carries responsibility when an automated system influences an outcome, rather than allowing accountability to disappear inside a technical process.

That principle reaches into public administration, employment, policing, finance and other areas where automated assessments can affect a person’s rights or opportunities. The source resolution does not list specific sectors or prescribe a single oversight model. It does require states to consider human rights when they write laws and regulatory frameworks.

The Council also calls on governments to combat discrimination and other harmful effects. That places attention on the data used to train systems, the criteria used to assess people and the procedures available when an AI-supported decision causes harm.

The wording leaves implementation to national authorities. Switzerland’s federal government, cantonal administrations and regulators will need to assess how the principle fits with existing duties of transparency, due process and personal data protection. The Geneva resolution supplies a direction, while detailed safeguards remain a matter for lawmakers and institutions.

The resolution links AI governance to participation, infrastructure and development. It says all countries, including developing nations, should have a voice in discussions over how the technology is regulated. That demand reflects the global reach of systems often designed and controlled by a relatively small number of governments and companies.

The Council also urges stronger international cooperation and greater coherence between existing initiatives. States, the UN and other stakeholders should promote digital public goods, with particular attention to access to digital infrastructure.

The language broadens the debate beyond the conduct of individual companies. It points to the conditions that determine who can build AI, who can inspect it, who can benefit from it and who bears the risks when systems fail. Without access to infrastructure and technical expertise, participation in global rule-making can remain largely formal.

The resolution does not specify funding levels, institutional arrangements or a timetable for closing infrastructure gaps. Those details will matter as governments turn the consensus text into policy. Geneva’s diplomatic role will include making sure that participation remains wider than the states and companies with the greatest computing capacity.