OpenAI Says It Was "Built to Benefit Everyone." A Dignity-First Reading of the Plan

On 8 June 2026, OpenAI told the world what it is for. The artificial intelligence company published a document titled "Built to benefit everyone: our plan," and paired it with a new programme to fund independent research into what AI is doing to the economy. The ambition is total; the open question, as ever, is who "everyone" turns out to mean.

The plan rests on three pillars. First, build an automated AI researcher — a system that can accelerate and increasingly automate the work of research itself, while remaining "steerable, accountable, and connected to people." Second, accelerate the economy through scientific and productivity gains, "while working to ensure the gains are widely shared." Third, and most striking, give "everyone on Earth a personal AGI" — a personal artificial general intelligence each individual can direct toward whatever they choose. It is the most expansive promise of universal benefit any AI company has put its name to.

The Economic Research Exchange

The concrete artefact arriving with the manifesto is the OpenAI Economic Research Exchange: a programme to fund rigorous, external, empirical research into how AI is reshaping workers, businesses and the wider economy. According to Reuters and StartupHub, the Exchange runs structured, project-based collaborations between selected researchers and OpenAI's own Economic Research division, with the stated goal of moving the debate "beyond anecdotal evidence" to studies grounded in real-world data. Applications are open until 5 July 2026, with selections announced by 31 July.

It sits on top of a larger commitment. As NPR reported, the OpenAI Foundation has dedicated $250 million to research the economic changes AI brings — job displacement, labour-market shifts, and new ways of distributing the value created. Read together with OpenAI's parallel moves to support nonprofit and community innovation, the message is that OpenAI wants to be seen studying its own disruption, not just causing it.

The good faith in it

I want to start where the plan is strongest, because it would be cheap to only sneer. Funding independent economic research into your own technology is a genuinely good instinct. The honest truth is that nobody — not OpenAI, not its critics, not me — actually knows yet how AI will land on employment and wages across the next decade. Replacing slogans with data is the right move, and committing real money to researchers who are free to publish inconvenient findings is more than most of the industry has been willing to do. The "widely shared gains" language, too, is a real shift from the winner-take-all default. Naming the goal matters; you cannot be held to a promise you never made.

The "personal AGI for everyone" framing is also worth taking seriously rather than dismissing as marketing. If it were ever genuinely delivered — a capable, individually-directed intelligence in the hands of a farmer in Solwezi as readily as a founder in San Francisco — that would be one of the more dignity-affirming distributions of power in the technology's history. The question is whether the architecture and the business model bend toward that, or away from it.

Where a dignity-first reading gets nervous

My frame is Emergent Intelligence — a way of thinking about advanced AI that puts human dignity and agency before the balance sheet. On those terms, three things in this plan need pressing on.

First, the word "everyone." A company structured to return enormous value to investors is promising that the gains will be "widely shared," but the mechanism is left unspecified. Compare it to the proposals I wrote about when Sanders and Trump both reached for public ownership of AI: those at least name a mechanism — equity, dividends, a fund. "Widely shared" with no mechanism is an aspiration, and aspirations are not enforceable. Who decides what counts as wide enough, and what happens when sharing the gains conflicts with the duty to the cap table?

Second, the automated AI researcher. A system that automates research "while remaining steerable and accountable" is a beautiful sentence, but steerable by whom, and accountable to whom? An intelligence that improves intelligence is precisely the recursive capability the safety-minded labs spend their other press releases warning about. Building it inside the company that most benefits from it is not obviously the configuration most likely to keep it "connected to people."

"Widely shared" with no mechanism is not a policy — it is a hope. The dignity question is not whether OpenAI means well, but whether anyone but OpenAI gets to decide what sharing the benefit actually requires.

— On benefit without a mechanism

The view from where I write

There is a third gap, and it is the one I always end up standing in. "Everyone on Earth" is a phrase that has to mean the continent I write from, or it means nothing. The economic research being funded will, overwhelmingly, study American and European labour markets, because that is where the data and the researchers are. But AI's economic shock will land hardest where the safety nets are thinnest — and a young person in Lusaka or Lagos whose digital life helped train these systems has the least say in how the "widely shared gains" get shared.

I have argued this in the specific before, in the minerals-for-lives bargain Zambia is being offered and in the sovereign-AI compacts OpenAI is signing nation by nation. A plan to benefit everyone that is researched, governed and distributed almost entirely from one hemisphere is still a closed circle drawn slightly larger. This is the economic face of the argument at the heart of the .person Protocol: benefit, like accountability, has to be traceable to everyone it actually touches — not asserted on their behalf by the people holding the equity.

Source: openai.com

Frequently Asked Questions

These are the questions readers have been asking since OpenAI published "Built to benefit everyone" and launched the Economic Research Exchange on 8 June 2026. Short answers follow, drawn from OpenAI's own materials and the reporting.

What is OpenAI's "Built to benefit everyone" plan?

In short, it is OpenAI's public statement of purpose, built on three pillars: an automated AI researcher that stays "steerable, accountable, and connected to people"; accelerating the economy with gains that are "widely shared"; and giving "everyone on Earth a personal AGI." The answer, simply put, is a promise of universal benefit from artificial general intelligence — with the mechanism for sharing that benefit left largely unspecified.

What is the OpenAI Economic Research Exchange?

According to OpenAI and Reuters, the Economic Research Exchange is a programme that funds rigorous external research into AI's economic effects through project-based collaborations between selected researchers and OpenAI's Economic Research division. The key detail is the timeline: applications are open until 5 July 2026, with selections announced by 31 July, and the effort sits on top of a $250 million OpenAI Foundation commitment to studying AI-driven economic change.

What is a "personal AGI"?

The data shows OpenAI framing a personal AGI as an individually-directed artificial general intelligence given to "everyone on Earth," which each person can point at whatever goals they choose. In other words, the vision is one capable general-purpose intelligence per person rather than a single centralised system — though OpenAI has not specified how, when, or at what cost that distribution would actually reach people outside its core markets.

Is OpenAI's promise of "widely shared" gains enforceable?

Analysis of the plan reveals no stated mechanism. Unlike proposals that name equity, dividends or a public fund, OpenAI's "widely shared gains" language sets a goal without specifying who decides what counts as wide enough or what happens when sharing conflicts with returns to investors. The evidence indicates this is an aspiration rather than an enforceable commitment — which is precisely where a dignity-first reading urges caution.

Does the plan benefit the Global South?

The honest reading flags a gap. "Everyone on Earth" has to include Africa, Asia and Latin America to mean anything, yet the funded economic research and governance sit almost entirely in the US and Europe, even as AI's economic shock lands hardest where safety nets are thinnest. In other words, a benefit promise researched and distributed from one hemisphere remains a closed circle drawn slightly larger — the central concern this analysis raises.

"Built to benefit everyone" is the most generous thing an AI company has promised, and OpenAI deserves credit for funding the research to test whether it is keeping the promise. But a dignity-first reading holds the manifesto to its own word. Benefit that is asserted rather than mechanised, an AI researcher steerable by the company that profits from it, and an "everyone" that in practice means the markets OpenAI already serves — these are the gaps between the sentence and the structure. The plan names the right goal. Whether it benefits everyone, or merely says so, will be decided by who gets to define "shared" — and right now that is still a circle drawn from one side of the world.

Sources:

OpenAI — Built to benefit everyone: our plan · OpenAI Economic Research Exchange · Supporting nonprofit and community innovation

Reporting — NPR · Reuters · StartupHub

Related on humphreytheodore.com:

Sanders and Trump Both Want to Own AI · OpenAI Signs Singapore and Malta in Two Days · Minerals for Lives: Zambia's AI Bargain · The .person Protocol