Marc Benioff, Jensen Huang, and World Leaders Launch UN's AI for Good Global Commission

For years, the gap between the people building transformative AI and the people governing it has been one of the defining tensions in technology policy. On July 1, 2026, the United Nations made the most direct attempt yet to close it: the launch of the AI for Good Global Commission, a body specifically designed to put the world’s most powerful technology executives alongside heads of state in a single, action-oriented governance structure.

The commission will hold its inaugural meeting on July 8 in Geneva, during the ITU’s AI for Good Global Summit — just days after the UN’s Global Dialogue on AI Governance concludes on July 7.

Who’s in the Room

The commission’s membership reads like a dispatch from the intersection of tech power and political authority.

Marc Benioff, CEO of Salesforce — who has spent the past several years positioning himself as the AI industry’s most vocal advocate for governance guardrails — will co-chair alongside Paul Kagame, the Rwandan president whose country has become an unlikely model for AI-forward national development policy in the Global South. ITU Secretary-General Doreen Bogdan-Martin will serve as permanent vice-chair, providing the institutional anchor for a body that needs to produce durable outputs rather than ceremonial communiqués.

The executive roster spans the full depth of the AI supply chain. Jensen Huang, NVIDIA’s founder and CEO, whose chips underpin virtually all frontier AI training, is a member. Andy Jassy, Amazon CEO, brings the perspective of the world’s largest cloud infrastructure provider. Brad Smith, Microsoft’s president, represents the company that has made the largest single investment in any AI lab. Anthropic co-founder Jack Clark and Cohere co-founder Aidan Gomez provide the frontier lab perspective. Estonian President Alar Karis — whose country built one of the world’s most sophisticated digital government systems — joins alongside policymakers from Kazakhstan, Namibia, Nigeria, Saudi Arabia, and Singapore.

The geographic spread is deliberate. One of the core critiques of AI governance efforts to date is that they have been dominated by the perspectives of wealthy democracies — the EU’s AI Act, the OECD’s AI Principles, the Bletchley Park declaration — while the countries most likely to be shaped by AI without shaping it have been present mainly as audiences. This commission’s design explicitly includes Global South voices at the same level as Silicon Valley executives.

What It’s Designed to Do

The commission is deliberately structured to be different from the Global Dialogue on AI Governance, which runs for two days immediately beforehand in Geneva and consists primarily of government diplomats negotiating international frameworks. The Dialogue is the traditional UN mode: slow, representative, consensus-seeking, and often productive only after years.

The AI for Good commission is meant to be faster. Its stated design principle is to be a “smaller, faster team of business executives” who can identify specific problems and move toward solutions without the procedural weight of full diplomatic consensus. In practical terms, this means it can publish recommendations, convene working groups, and broker voluntary commitments among its members — activities that don’t require ratification by the 193-member UN General Assembly.

The mandate is broad but centered on a specific question: how do you ensure that the most powerful technology in human history is used to help humanity rather than divide it? That phrasing appears in commission materials not as aspiration but as an operational framing — an acknowledgment that the risk is real and that the answer requires both technical expertise and political will.

The Governance Architecture

The ITU’s role in hosting and anchoring the commission matters. The International Telecommunication Union has been a forum for international technology coordination since 1865, and its existing structures — technical committees, observer processes, national membership — give the commission a procedural home that doesn’t need to be built from scratch.

More importantly, the ITU’s membership includes virtually every country on earth, including China, Russia, and the Gulf states whose AI governance relationships with the West remain complicated. A commission launched under UN auspices with ITU participation has a legitimacy claim that purely Western-initiated bodies do not.

The commission is structured to meet regularly, with the Geneva inaugural session to be followed by quarterly convenings that produce public outputs — reports, recommendations, and voluntary commitments. Members are expected to show up, not just lend their names. Whether that expectation holds for executives managing companies worth hundreds of billions of dollars under the daily scrutiny of AI deployment decisions remains to be seen.

The Timing Is Not Accidental

The commission is launching in the middle of the most consequential month in AI policy since ChatGPT was released. The same week, GPT-5.6 Sol is being previewed under U.S. government review. The EU’s AI Act high-risk obligations are moving toward August enforcement. The U.S. Congress is actively negotiating federal AI preemption legislation. Anthropic’s Fable 5 has just returned from a government-mandated export ban.

In this environment, a body that can convene the CEO of NVIDIA and the President of Rwanda in the same room is not just symbolically significant. It is potentially the only governance mechanism with the composition to address problems that are simultaneously technical, commercial, geopolitical, and developmental.

Benioff, speaking at the commission’s launch, framed the challenge in terms that have characterized his recent public positions on AI: the technology is arriving faster than any single governance system can process, and the choice is not between moving fast and governing responsibly — it is between governing proactively and governing reactively after damage is done. The commission, in his framing, is an attempt at proactive governance by the people with the most ability to shape outcomes.

What Success Looks Like

Commission governance bodies fail in predictable ways: they produce elegant reports that nobody implements, they fracture along geopolitical lines when specifics are required, or they lose momentum when busy members’ attention moves elsewhere.

The early test of whether this commission is different will come from its first six months. If the July 8 inaugural meeting produces a specific set of commitments — measurable targets on AI safety testing, concrete mechanisms for sharing technical capacity with developing countries, or a defined timeline for working principles on AI in critical infrastructure — that would signal a body capable of producing durable outputs.

If it produces another declaration of principles without implementation architecture, it will join the long list of high-profile AI governance initiatives that demonstrated the seriousness of the problem and the limits of voluntary coordination.

The AI for Good commission’s membership is unprecedented. What it does with that membership will determine whether it represents a genuine inflection point in AI governance or another iteration of an approach that has repeatedly fallen short of the pace of AI development.

The world will be watching Geneva on July 8.