A president and a CEO at the same table — the UN and ITU stand up an 'AI for Good' commission

On July 2, 2026, Geneva produced an unusual picture. A sitting head of state and the CEO of a major U.S. tech company sat side by side, each with a co-chair nameplate in front of them. That's the day the International Telecommunication Union (ITU) formally launched the "AI for Good Global Commission." The co-chairs are Rwandan President Paul Kagame and Salesforce CEO Marc Benioff, with ITU Secretary-General Doreen Bogdan-Martin serving as vice-chair. More than 40 founding members are on the roster — heads of state, government ministers, Big Tech CEOs, and heads of UN agencies, all on one list.

The core message is simple. Don't let AI harden into a technology that only fattens a handful of rich countries and a few giant companies. One of the goals the commission put front and center is closing the "digital divide" — the 2.2 billion people still with no internet access at all — so the AI era doesn't widen that gap even further. At the launch, Bogdan-Martin drew a hard line: "No organization can single-handedly put AI at the service of all humanity." Kagame added, "Technology is supposed to be a force for good, and we have a responsibility to use it accordingly."

But right here the doubts start creeping in. An international commission on AI governance isn't exactly new. Inside the UN alone, several tracks are already running. And most bodies like this issue "recommendations" but can't "enforce" anything. There's a long history of presidents and CEOs taking a photo, dropping a slick declaration, and calling it a day. So in this piece I'll try to unpack — as evenly as I can — who's angling for what, what could actually change and what won't, and how this collides with the other AI governance games running in parallel right next door. To give away the ending a little: whether this really moves the board is genuinely too early to call.

Kagame, Benioff, Bogdan-Martin — and the 40-plus founding members

Start with the people at the table. This commission has three faces. First, co-chair Paul Kagame, President of Rwanda. Rwanda is known for pushing digital, drone, and fintech tech as national strategy, and putting Kagame up front is a deliberate move to signal that "this commission isn't a Silicon Valley party — it carries developing-world voices too." Whenever the AI-gap conversation happens in the Global South, Africa comes up, and one president is now shouldering a lot of that representation.

Second, co-chair Marc Benioff, CEO of Salesforce. Honestly, this pairing is the real headline. A sitting U.S. Big Tech CEO taking a co-chair seat at a commission created by a UN agency is not an everyday thing. Salesforce is betting its future on AI-agent products like Agentforce, so for Benioff, sitting dead center at the table where global AI norms get drafted matters commercially, not just symbolically. And that's exactly where the conflict-of-interest question comes in: is it right for a party who'll be subject to the rules to co-chair the body writing them?

Third, vice-chair Doreen Bogdan-Martin, ITU Secretary-General. ITU is an old UN specialized agency that historically handled telephone, spectrum, and telecom standards, but over the past few years it has grown an annual "AI for Good" summit and dramatically raised its profile in AI governance. Bogdan-Martin sits at the center of that shift, and this commission is bolted on top of that summit brand. In other words, it's less a brand-new organization than an existing "AI for Good" ecosystem given extra weight by attaching top-tier figures to it.

Then there are the 40-plus founding members, and the roster is glossy. On the Big Tech side you'll find names like Amazon CEO Andy Jassy, NVIDIA CEO Jensen Huang, Microsoft Vice Chair Brad Smith, Anthropic's Jack Clark, and Cohere CEO Aidan Gomez. The corporate lineup — Salesforce, Microsoft, Google, Amazon, NVIDIA, Qualcomm, Accenture, Anthropic, Vodafone, Orange, Grab, MTN Group, ZTE, and Cohere — blends Western, African, and Asian firms. On the international-org side, the African Union, UNESCO, the World Trade Organization (WTO), and the World Intellectual Property Organization (WIPO) are attached, and among national representatives you have Estonia's President Alar Karis alongside policymakers from Kazakhstan, Namibia, Saudi Arabia, Singapore, and Nigeria. This "all-star" mix — developed and developing, Western and non-Western, government and company and international body all in one list — is the commission's core selling point.

What actually happened — the skeleton and the schedule

To be precise, what happened on July 2 was closer to an "org chart reveal." The mission, the leadership, and the founding-member list went public; the actual work hasn't started yet. The direction the commission set out reads in roughly three strands: (1) strengthening trust in AI, (2) supporting responsible innovation, and (3) spreading AI's economic and social benefits broadly. The practical task cutting across all three is that digital-divide goal — reaching the 2.2 billion who are still offline. Focusing on "access" and "benefit distribution" rather than another abstract ethics declaration is, if anything, this commission's distinguishing color.

The most important next event is just days away. The commission's inaugural meeting runs during the "AI for Good Global Summit 2026," held July 7-10 in Geneva. That summit is ITU's big annual event, and this year it's folded into a "Digital Week" running July 6-10. So the launch announcement dropped early on July 2, while the actual first meeting takes place on the summit stage under the spotlight.

Here's a timing detail worth flagging. In the same Geneva, on almost the same dates, a completely separate track — the "UN Global Dialogue on AI Governance" — holds its first session July 6-7 at Palexpo. That one is an intergovernmental official platform created by the UN General Assembly under the Global Digital Compact. Which means early-July Geneva is essentially AI-governance events stacked on top of each other within a few days, and this commission is best understood as one card on that crowded stage.

Item Detail
Official name AI for Good Global Commission
Launch date July 2, 2026 (Geneva)
Co-chairs Paul Kagame (President of Rwanda) · Marc Benioff (CEO of Salesforce)
Vice-chair Doreen Bogdan-Martin (ITU Secretary-General)
Founding members 40+ (heads of state, ministers, Big Tech CEOs, UN agency heads)
Core goals Strengthen trust in AI · support responsible innovation · broaden benefit distribution · close the 2.2B digital divide
Inaugural meeting AI for Good Global Summit 2026 (July 7-10, Geneva)
Nature Advisory/recommendation-focused; no legal binding force

What each side gains

With a setup like this, the picture only sharpens once you ask "who takes home what." Start with ITU and Bogdan-Martin. For ITU, this commission is a card to stake a claim as "the real hub of AI governance inside the UN." As noted, several UN tracks touch AI, and assembling top-tier leaders and Big Tech CEOs on one board is a way for ITU to sharply raise its standing in that competition. It also helps rebrand a somewhat old-fashioned telecom-standards body into "the coordinator of the AI era."

What Kagame and the developing-world camp gain is the seat at the table itself. AI-norm debates have long been led by tech powers like the U.S., China, and the EU, with the Global South notified after the rules were already written. Having an African head of state hold a co-chair seat is symbolically significant. Whether it stays symbolic, or turns into tangible outcomes like resources, infrastructure, and technology transfer, is an entirely separate question.

What Benioff and Big Tech gain is a cleaner calculation. First, they secured a place to help design the norms from inside the room rather than being regulated from outside. Second, the halo of "partnering with the UN to build good AI." As public distrust of AI and regulatory pressure grow, voluntarily participating in governance is a valuable shield for a company. Of course, critics zero in on exactly this point: when the party to be regulated co-chairs the design of that regulation, the outcome tends to tilt toward "voluntary compliance" over "enforceable rules."

And ordinary people? Honestly, at this stage everything comes with an "if it works" attached. If the commission genuinely lifts AI access in lower-income countries and pushes public uses like multilingual support, education, and healthcare, benefits can flow downward. But that only holds when the commission moves past declarations into budgets and execution — and there's no evidence of that yet.

We've seen this before — global AI-governance wins and failures

The most useful thing when looking at a body like this is to line it up next to past cases. On the success side, there's UNESCO's Recommendation on the Ethics of AI (2021). Adopted by 193 member states, it was the first global AI-ethics standard, and it later became a real reference frame in many countries' domestic AI-policy debates. It had no binding force, but it created a "common language," and that mattered. The OECD AI Principles (2019) are similar: not law, but once the G20 embraced them they became the conceptual foundation for later national regulation.

On the flip side, there are plenty of "declared and then fizzled" cases. Joint statements on AI from various international summits are the classic example — the fine phrasing and the photos remained, but with no follow-up implementation mechanism, a year or two later almost no one remembered them. The problem is always the same: no means to enforce, no budget attached, and blurry accountability over who does what by when. However glossy the roster, it ends as a promise on paper.

"Voluntary governance with heavy Big Tech involvement," in particular, has a mixed report card. Companies making voluntary safety pledges is fast and flexible for early norm-setting, but it has tended to retreat or loosen the moment real interests are on the line. The critique that self-regulation gets used as a "delay device" against actual regulatory legislation is a well-worn refrain, too. For this commission to dodge that trap, it ultimately needs to build a bridge that translates an advisory body's recommendations into the "actual rules" of an intergovernmental official track like the UN Global Dialogue.

The lesson of history, then, is this: global AI commissions have often succeeded at "creating a common frame" but almost always failed at "producing enforceable outcomes." Which way this commission goes will be decided not by the gloss of the roster, but by the implementation documents and budgets that show up a few months from now.

The neighboring tracks — the UN Global Dialogue, and the overlapping games

This part is actually the most interesting angle in the whole story. The AI for Good Commission is by no means an isolated island. Right next door, another UN track — the "Global Dialogue on AI Governance" — is up and running. It's an intergovernmental official platform the UN General Assembly established under the Global Digital Compact, and its first session runs July 6-7 at Geneva's Palexpo. It's a venue where all governments and stakeholders gather to discuss international AI cooperation, share best practices, and take on human-rights, safety, and divide questions. A second session follows in New York in May 2027.

The difference in character between the two tracks matters. The Global Dialogue is a formal diplomatic stage oriented toward "government-led, General-Assembly-based, state-to-state consensus." The AI for Good Commission, by contrast, is closer to a hybrid advisory group blending "heads of state + CEOs + international-org heads." One is the orthodox multilateral line; the other is an execution-minded, notable-driven line. In theory they can be complementary — the commission pulls in ideas and resources, the Dialogue formalizes them into intergovernmental consensus. But in reality, there's plenty of room for a quiet contest over stage, budget, and attention, over "who is the face of AI governance."

Layer on the non-UN tracks too. Actual legislation lines like the EU's AI Act, the U.S.-led safety-institute network, the G7's Hiroshima Process, national AI safety institutes — the AI-governance terrain is already stacked and crowded. In that setting, a cold reading of one more commission is: "this didn't add a board that coordinates, it added one more board to be coordinated." Indeed, one of the biggest risks in AI governance is "fragmentation, where too many overlapping forums mean no one holds final responsibility."

So the commission's real test is "differentiation." It has to use the flexibility of an advisory body to do something the other tracks can't. For example, using the sheer power of putting presidents and CEOs in one room to connect actual investment to developing-world AI infrastructure, or to pull Big Tech's technology and data into public uses — an "execution bridge" role. Pull that off, and it creates synergy with the other tracks; fail, and it just adds a name to a crowded list as "one more commission."

So what changes — persona by persona

From the developing world's viewpoint, what changes in the short term is the "seat" and the "attention." Having your president or minister sit at the AI-norms table can be a starting point for negotiating leverage. But the real change comes when that seat leads to tangible programs like infrastructure investment, technology transfer, and talent development. Right now only the "frame" of that promise exists, and actual allocation depends on years of execution ahead. Coldly put, the symbolic progress is real, but the material progress is still undecided.

For AI companies, a channel opened to get involved in norm design from the very start. That's double-edged. Used well, it lets them help build sensible, predictable rules; used badly, it can look like a lobbying window that dulls regulation. Participating firms like Salesforce gain a "responsible-AI leader" image, but they also draw the watchful eye that asks, "if the conclusion tilts toward voluntary compliance, isn't that ultimately favorable to the companies?"

For policymakers and regulators, the "coordination burden" grows. Drafting domestic AI policy now means watching multiple signals at once — the EU AI Act, the UN Global Dialogue, and this commission's recommendations. Aligned well, international coherence improves; misaligned, you get confusion over "which beat do we march to?" So national regulators will watch closely how well this commission's output aligns with the UN's official track.

For ordinary users, there's almost no change you'll feel immediately. The effects of a body like this flow slowly and indirectly. At best, over several years, they can show up as safer AI defaults, broader multilingual and low-cost access, and responsible AI use in public services. But again, that's only if the commission converts declarations into execution — and at this point it's too early to say "this is how your life changes."

In one line: this launch is clear progress on the direction of "don't let AI stay the property of a few," but whether that direction reaches its destination remains a homework problem that future execution, budgets, and implementation mechanisms have to prove.

🥄 Three Things You're Probably Wondering

— Does this commission actually have the power to enforce anything? Honestly, no. It's an advisory, recommendation-oriented body, not a regulator that enforces rules. It has no legal binding force, and for its recommendations to become actual rules they'd have to be translated into an intergovernmental official track like the UN Global Dialogue or into national legislation. So it's too early to declare "the commission changes the world" — its real influence is decided by how much those recommendations feed into actual institutions, and that bridge hasn't been built yet.

— Isn't a Big Tech CEO co-chairing it a problem? That's exactly the central controversy here. With a party to be regulated sitting at the head of the table designing the regulation, critics warn the outcome risks tilting toward voluntary compliance over enforceable rules. On the other side, defenders argue norms only work if people who understand industry reality help write them. Which view is right will be proven by whether the commission can actually issue recommendations that could disadvantage companies — and that's still a wait-and-see phase.

— The UN already has several AI-governance venues, so why build another? This is the question the cold-eyed crowd asks most. Right next door the UN Global Dialogue (July 6-7, Geneva) runs as an intergovernmental official track, and adding the commission on top raises the fragmentation worry that "there's now one more board to coordinate." ITU's logic is that "we're the only ones who can put presidents and CEOs in one room as an execution-capable advisory group." If that differentiation lands, it's synergy; if not, it just adds one more name to a crowded list. You'll only know from the output a few months out.

References

Numbers and criteria are as of announcement and may change.