Summary

  • A US proposal dated 19 October would retitle ITU Resolution 214 around “super intelligence” and confine the Union's work to its telecommunications and ICT remit. Its new operative line would exclude work supporting policy, legal, governance, regulatory or ethics frameworks.
  • The text adds Council permission gates for new super-intelligence initiatives, including projects through AI for Good, and asks that platform to use extra-budgetary funding. Delegates at the November conference will decide what survives.

The American edit gives the ITU's AI desk a grander label and a smaller set of keys. Across its proposed revision of Resolution 214, “artificial intelligence” becomes “super intelligence”. Then comes the lock: ITU would be forbidden from supporting the development of policy or legal frameworks, governance, regulation or ethics in this field. Studying how the technology works in networks stays on the desk. Writing society's rules for it goes behind glass.

The 2022 resolution already limited AI work to ITU's mandate and core competence. It encouraged studies, information sharing, capacity building and an AI-ready telecom ecosystem, with annual reports to Council. The US adds an express exclusion. That sentence would give members a direct test when an attractive new initiative claims a place under the ITU name: is this telecommunications work, or a bid to write the policy around it?

The proposed permission slips are unusually specific. Council approval would precede any super-intelligence initiative, advisory group or project through AI for Good. A second clause asks Council approval for comparable work and partnerships outside the ITU study groups. Bureau directors would report SI activity to their sector advisory groups each year and seek approval for Secretariat-led initiatives outside those groups. The proposal also directs AI for Good towards extra-budgetary funding, including voluntary money for associated staff time, and towards deployment and adoption of SI technology. The public register records it as a conference proposal, not a change already in force.

Another file on the same conference table points elsewhere. China, India and Viet Nam's contribution 89 keeps the language of AI and asks the ITU to promote standards, guidelines and good practice for safe, secure and trustworthy AI applications in telecoms, along with skills work. Its ITU register lists the same November meeting. One proposal expands a practical standards assignment; the US text draws a hard line around governance work and moves new projects toward Council sign-off. The conference has two different red pens in front of it.

There is one important blank in the American text: it never defines the point at which AI becomes “super intelligence”. That matters because the proposed ban and approval gates use that term. Doha's final wording will determine whether the lock has a clear keyhole. Until then, the sharpest fact in the file is the proposed shift in authority: technical study with the ITU, new initiatives before Council, policy and regulation outside this resolution's remit.

Official sources