AI Rules Can Soon Go Global: The First International Treaty Is Emerging

The global AI governance landscape is changing.On 17 May 2024, the Council of Europe adopted the Framework Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law (CETS 225), a milestone agreement signed by Switzerland, the EU, UK, and others. When it opened for signature on 5 September 2024, it marked the birth of the first-ever binding international treaty on AI, requiring signatories to uphold sweeping legal obligations for AI transparency, fairness, accountability, and user protection.It will only become legally binding once five signatories, including at least three Council of Europe member states ratify or approve it.

From that point, a three-month countdown begins before it takes effect on the first day of the following month. As of mid‑June 2025, many countries have signed, but none have yet ratified, meaning the activation clock has yet to start.

‍A New Legal Frontier for AI

‍Unlike the EU AI Act, which focuses primarily on “high-risk” systems, this treaty takes a full-lifecycle, risk-based approach to AI governance, covering any AI system used by public authorities or by private actors acting on their behalf. The scope is broad, and the expectations are clear: align AI with democratic values and fundamental rights, or face regulatory consequences.

‍Here’s what the treaty demands:

• Lifecycle Accountability: AI systems must be governed from design to deployment, with continuous oversight and adaptability as risks evolve.

• Foundational Principles: Systems must respect human dignity, autonomy, transparency, non-discrimination, privacy, and the rule of law.

• Transparency and Redress: Users must know when they’re engaging with AI (rather than a human), and be empowered to challenge decisions and access effective remedies.

• Thorough Risk Management: Risk and impact assessments must be carried out early and regularly, with stakeholder participation and documented safeguards.

• Governance Capacity: States must be able to suspend or ban AI systems if their use poses unacceptable risks.

• Collective Oversight: A dedicated Conference of the Parties will coordinate implementation, promote digital literacy, and support inclusion of vulnerable groups.

This is not just a guideline, it’s a treaty. And once it enters into force, signatories will be legally obliged to embed these standards into national law and institutional practice.

‍And Once Ratified…If the fifth eligible country were to ratify and deposit its instrument on, say, 10 September 2025, the treaty would become legally active by 1 January 2026.That’s not far off and for organisations building or supplying AI systems in regulated sectors, the clock is effectively already ticking.The Convention underscores a broader reality: AI governance is no longer abstract, it’s accelerating, formalising, and globalising.

Between the EU AI Act, national strategies, and now this international treaty, the direction is clear: rules are multiplying, and compliance is becoming multidimensional. Public and private actors alike must demonstrate that their AI systems are not only technically robust, but also legally and ethically sound.That requires a proactive, structured approach, one that anticipates legal developments, embeds accountability, and integrates rights-based design from the ground up.

At Novius Consulting, we help organisations interpret, anticipate, and adapt to this evolving AI regulatory environment.Because navigating AI law isn’t just about risk avoidance, it’s about building systems that are credible, resilient, and future-ready.