Can middle powers set AI rules without superpower buy-in?

Global Coverage Synthesis

Twenty nations back human-control AI pledge; US, China absent

Can middle powers set AI rules without superpower buy-in?

At the UN’s High-level Week, 20 governments endorsed keeping frontier AI under human direction and floated an international watchdog to set and enforce standards, even as several European heavyweights and both the U.S. and China withheld signatures.

Story Summary

On the sidelines of the UN General Assembly, leaders from more than 20 countries and the European Commission president backed a declaration to keep frontier AI under human direction, oversight and control, with signatories also floating a new international body empowered to set and enforce standards—while the United States, China, and several major European powers declined to sign. The move tests whether middle powers can shape binding governance and keep multilateralism relevant amid great‑power rivalry, with consequences for global safety, security, and who sets the rules for the Global South. The unresolved question is both who leads and what model prevails—UN‑anchored regulation, a new institution, or U.S.-led norms—against clashing risk framings that range from a nuclear‑style non‑proliferation push to “pacing” frontier capabilities and China‑focused threat doctrines.

Full Story

Twenty countries back human‑control pledge and propose global AI oversight as U.S. and China hold back

Narrative Snapshot

European and middle‑power outlets center the push to assert agency outside the U.S.–China rivalry. Le Monde and RT both highlight that more than twenty governments, including Germany and Canada, endorsed a call to keep frontier AI “under human direction, oversight and control,” but that the U.S., China and several key European states declined to sign. Al Jazeera advances the story by noting signatories are proposing an international institution to set and enforce standards, signaling an appetite for rulemaking with teeth that aims to transcend ad hoc principles.

Coverage splits over locus of authority. The New York Times frames AI as a two‑power contest in which the United Nations risks irrelevance, while the Mail & Guardian argues that precisely because voluntary guardrails lag frontier systems, the UN must lead to protect the Global South and set credible standards. From the UK, the Guardian reports a party‑political call for a nuclear‑style non‑proliferation treaty and a global pause on super‑intelligent AI, amplifying the analogy that also appears in Canada’s Toronto Star, which urges international cooperation to rein in risks.

U.S. reporting is consumed by domestic positioning and China strategy. Fox News carries President Donald Trump’s dismissal of existential risk as a “hoax,” former President Barack Obama’s call for stronger federal regulation, and arguments to use Cold War‑era legal tools like the Defense Production Act if an AI system threatens national security. The same outlet also features industry voices urging “pacing” frontier capabilities and coordinated democratic standards, alongside hawkish warnings that China leverages open‑source releases and model replication to erode U.S. advantage. A CGTN commentary reframes the rivalry around structural industrial differences—software‑heavy in the U.S. versus more embodied AI in China—implying divergent incentives in governance.

What Happened

On the sidelines of the UN General Assembly’s High‑level Week in New York, leaders from 20 countries, joined by the European Commission president, endorsed a declaration urging that frontier AI remain under human direction, oversight and control. RT reports the effort was spearheaded by Finland’s president and Norway’s prime minister, remains open to additional endorsements, and included Germany, Spain and smaller EU states; Le Monde notes Canada was among supporters. The United States, China and several major European powers, including the UK, France and Italy, did not sign, according to RT and Le Monde. Al Jazeera adds that backers floated creating an international oversight body empowered to set and enforce standards. In parallel, UK Liberal Democrat leader Ed Davey called for a global non‑proliferation treaty and a pause on super‑intelligent AI, while OpenAI, cited by Clarin, urged U.S. leadership on global norms and safety standards. U.S. coverage detailed partisan divides, emergency authorities, and industry proposals to slow capability gains.

Why It Matters

The cluster of proposals—human‑control commitments, an international oversight body, and a non‑proliferation framing—tests whether middle powers can shape governance when the two leading AI states stay outside the room. RT and Le Monde’s reporting that the U.S. and China withheld signatures underscores the core capacity challenge identified by the New York Times: multilateral relevance amid great‑power competition. Al Jazeera’s account of a standards‑setting institution and the Mail & Guardian’s case for UN leadership point to institutional choices between UN‑anchored arrangements and coalitions of the willing. Domestic political divisions in the U.S., captured by Fox News’ coverage of Trump’s dismissal of existential risk and Obama’s call for stronger federal regulation, bear directly on Washington’s ability to coordinate internationally or exercise the leadership urged by OpenAI in Clarin. CGTN’s depiction of structurally different AI industries suggests governance preferences may also diverge by sectoral makeup, complicating consensus on scope and enforcement.

Diverging Narratives

Who leads and on what basis remains unsettled. Clarin reports OpenAI arguing for U.S. leadership to set global benchmarks and safety standards, while the Mail & Guardian advocates a UN‑led regime to protect the Global South, and Al Jazeera describes signatories exploring an international body with enforcement powers. Le Monde and RT emphasize that the principal AI powers and some European heavyweights did not sign, reinforcing the New York Times’ characterization of a two‑power race that sidelines broader multilateralism.

Risk framing diverges sharply. The Guardian relays a call for a nuclear‑style non‑proliferation treaty and a pause on super‑intelligent AI, aligning with the Toronto Star’s nuclear analogy. Fox News, by contrast, reports President Trump labeling existential‑risk concerns a “hoax,” even as the outlet features countervailing industry views from Anthropic’s CEO advocating to slow capability advances and coordinate democratic safety standards. Fox News also elevates a threat model focused on Chinese exploitation of open‑source systems and model replication, with a research fellow quoted as seeing a “multi‑front war,” diverging from simple race metaphors. A CGTN commentary shifts the lens to underlying industrial structures—software versus embodied systems—suggesting governance debates may be downstream of different innovation bases.

What Happens Next

Whether the 20‑nation declaration matures into an institutional design effort will hinge on follow‑through toward the international oversight body Al Jazeera reports was proposed; signals would include a scoping process, a mandate draft, or alignment with existing UN organs as urged by the Mail & Guardian. The declaration’s openness to new endorsements, noted by RT, creates a test of coalition‑building: watch for uptake by major European nonsignatories or engagement from the U.S. and China, which Le Monde and RT say stayed out.

In the U.S., Fox News highlights several decision points with external effects: potential use or lapse of Defense Production Act authorities set to expire in December, choices on export policy for advanced AI chips raised in an op‑ed, and whether policymakers respond to calls, reported by Fox News, to slow capability gains through embedded evaluators and coordinated standards. In the UK, the Guardian’s reporting on a proposed AI non‑proliferation treaty frames a domestic debate whose resonance with government policy will be visible in future conference statements or legislative proposals.

How This Story Was Built

EDITORIAL METHOD

This page is a synthesis generated from cross-source coverage, then reviewed and published as a standalone narrative.

SOURCES

15 sources analyzed

OUTLETS

10 distinct publishers

COUNTRIES

9 source countries

DIVERSITY SCORE

84% (very high)

Show full editorial details

SOURCE TIMELINE

Coverage window from 16 Sep 2026 to 22 Sep 2026.

OUTLETS LIST

Al Jazeera English, CGTN, Clarin, Fox News, Le Monde, Mail & Guardian, New York Times, RT (Russia Today), The Guardian, Toronto Star

COUNTRIES LIST

Argentina, Canada, China, France, Qatar, Russia, South Africa, USA, United Kingdom

SOURCE MIX

3 ownership types 2 media formats 6 source regions

DIVERSITY NOTE

This score estimates how varied the source set is across outlets, countries, ownership and media formats. Higher means broader source diversity.

TRACEABILITY

All source links are listed below for verification.

PUBLICATION

Editorial review completed and published on 23 Sep 2026.

Listed from newest to oldest source publication.

Sources Analyzed

How to Cite This Story

Nereid Atlas Editorial Desk. "Twenty nations back human-control AI pledge; US, China absent." Nereid Atlas, . <https://www.nereidatlas.com/stories/2026-09-23-can-middle-powers-set-ai-rules-without-superpower-buy-in>