On 5 September 2024 the Council of Europe opened its Framework Convention on Artificial Intelligence for signature. It is the first legally binding international AI treaty, grounded in human rights, democracy and the rule of law, and consistent with the EU AI Act and other existing rules. The European Union has signed. For AI companies whose main market is Europe, the significance is not an immediate new penalty regime but the shift of AI governance from policy advocacy into an international obligation that can be invoked.
Where the binding force sits
The convention is framework-based: parties commit to adopting measures in domestic law consistent with its objectives, covering activities across the AI lifecycle, with particular attention to use by public authorities. It does not replace regional legislation such as the EU AI Act; it requires domestic regimes to align with its principles. Direct litigation risk for companies is limited. The indirect effect matters more — once a state transposes the obligations, the conditions for market access, public procurement and regulatory scrutiny change.
Public-sector use is the first observation window
The convention gives prominence to AI use by public authorities. Procurement, public services, law enforcement and judicial support usually demand greater transparency, human oversight and avenues of redress. Companies supplying AI systems to public bodies will meet these requirements in tender documents, contract clauses and acceptance criteria. Writing the convention’s principles into product documentation and contract annexes in advance is more effective than assembling them during a bid.
Human rights due diligence moves from voluntary to routine
Because the convention rests on human rights, democracy and the rule of law, corporate human rights due diligence is no longer optional brand narrative. Identifying affected persons, assessing the severity of impact and establishing complaint and remediation mechanisms already appear in due diligence questionnaires from European customers and investors. For companies selling into Europe, a document explaining how a model affects users’ rights is becoming a precondition for cooperation.
Relationship with the EU AI Act
Consistency with the AI Act does not make the two interchangeable. The Act sets product-side obligations by risk tier; the convention provides principle-level constraints under international law. Companies in Europe face two logics at once: pre-market conformity assessment and documentation on one side, continuing human rights and governance commitments on the other. Treating them as the same thing tends to leave gaps in the documentation set.
Three preparations for entering the European market
Confirm where the product is used within signatory states, particularly in public authorities and regulated sectors. Check whether existing governance documents cover human rights impact assessment and human oversight. Allocate responsibility and update obligations in contracts for changes in regulation.
For rule mapping in the European market, see our EU AI Act service; for multi-country deployment and cross-border arrangements, see multi-jurisdiction compliance and GDPR cross-border.
This article is based on the Framework Convention on Artificial Intelligence opened for signature by the Council of Europe on 5 September 2024, and on public information. It is for reference only and does not constitute legal advice.