Alongside the national AI laws of 2026, the international machinery for governing AI took real shape. The United Nations stood up its first standing scientific body on AI and held its inaugural Global Dialogue, India’s summit produced the New Delhi Declaration, and the EU ratified the world’s first binding AI treaty, though that treaty is not yet in force. None of this binds a developer directly, but it sets the frame that national rules increasingly follow.

What happened

The UN Independent International Scientific Panel on AI is the headline. The General Assembly appointed its 40 members on 12 February 2026, and the panel elected co-chairs Yoshua Bengio and Maria Ressa on 3 March. It released a Preliminary Report on 1 July 2026 and the inaugural UN Global Dialogue on AI Governance convened in Geneva on 6 to 7 July, alongside the ITU AI for Good summit. Both bodies were created by UN Resolution A/RES/79/325, adopted in August 2025, and are modeled loosely on the IPCC.

At the regional and multilateral level: India’s AI Impact Summit (16 to 21 February 2026) adopted the New Delhi Declaration on AI Impact on 19 February, endorsed by more than 85 countries and international organisations, as the successor to the Bletchley, Seoul, and Paris summits. The EU ratified the Council of Europe Framework Convention on Artificial Intelligence on 15 May 2026, the first binding international AI treaty, but as of mid-July it has only one ratification and has not met the threshold to enter into force. The OECD launched version 2.0 of its Hiroshima AI Process reporting framework on 28 May 2026, extending voluntary AI reporting to smaller companies.

Why it matters for builders

These bodies do not regulate you, but they steer who does. The UN panel is meant to become the shared evidence base that national regulators cite, much as the International AI Safety Report already is. The New Delhi Declaration reframes the global conversation around equitable access and impact rather than only safety, which shifts what future rules emphasise. And the Council of Europe treaty, once it enters force, will bind its parties to AI principles that flow down into national law.

The practical read: expect convergence over time on transparency, risk assessment, and documentation, because these international frameworks all point the same way, and national laws like the EU AI Act implement them. Building good governance now is a hedge against all of them.

Sources

Further reading