The DeepMind Institute (DMI) is a publishing forum on artificial general intelligence launched by Google and Google DeepMind on September 16, 2026. Its founding statement, written by its three directors, Shane Legg, James Manyika and Demis Hassabis, describes it as a platform for researchers and thinkers from across Google DeepMind, Google and the wider global research community to publish ideas about a world with AGI, and names Legg as managing editor (Introducing the DeepMind Institute (Legg, Manyika and Hassabis, September 2026)). Axios described Manyika as a Google senior vice president and Hassabis as DeepMind's chair, and characterized the Institute's purpose as giving researchers inside and outside Google a venue for essays and debate on the effects of AGI on society (Source: axios.com). It publishes at institute.deepmind.com under the description "Bold, well-evidenced thinking on AGI from the lab that pioneered the field" (Source: institute.deepmind.com).
Founding statement
The Institute's mission statement reads: "The DeepMind Institute aims to advance bold thinking about the safe development of artificial general intelligence (AGI), its beneficial use, and its implications for society" (Introducing the DeepMind Institute (Legg, Manyika and Hassabis, September 2026)).
The founding statement, "Introducing the DeepMind Institute," states that "the rapid pace of innovation suggests we're now approaching artificial general intelligence (AGI), a system that exhibits all the cognitive capabilities of the human brain," and that although AI still fails at some basic tasks and lacks the consistency and creativity of full AGI, "we expect those gaps to be closed soon." It presents AGI as potentially "the ultimate tool for accelerating scientific breakthroughs" in disease, clean energy and economic prosperity. It identifies cybersecurity and biorisks as current implications of more capable AI, together with "the potential for loss of control in future self-improving systems" (Introducing the DeepMind Institute (Legg, Manyika and Hassabis, September 2026)).
The statement frames the Institute's work around three questions: what society will value and how AGI will affect what it means to be human; how to build and govern AGI systems safely, along with "the communities of agents they will form"; and which institutions and policies society will need to adapt to AGI or reimagine. It states that the researchers and thinkers the Institute publishes, drawn from Google DeepMind, Google and the wider research community, "will not always agree, and they will likely change their minds, as more data and information comes to light at the fast-moving frontier," and that the Institute was created "because broad-based intellectual discussion and debate are required to arrive at a consensus about how to address the challenges and opportunities we face as a society" (Introducing the DeepMind Institute (Legg, Manyika and Hassabis, September 2026); Source: axios.com). It says answers about AGI's development "shouldn't be technologists alone who provide them," and that responsibility for shaping the new era belongs to society as a whole, "including the arts and humanities, and governments" (Introducing the DeepMind Institute (Legg, Manyika and Hassabis, September 2026)).
Organization
The Institute is a Google and Google DeepMind undertaking rather than an independent body; its site is served from a Google-operated domain under Google's privacy and terms policies. The founding statement lists Legg, Manyika and Hassabis as the "DeepMind Institute directors" and Legg as managing editor, which places editorial responsibility for the essays with him, and gives an editorial contact address at google.com. The site's disclaimer describes DMI as "a platform, started by researchers from Google and Google DeepMind," and states that its pieces "are intended to serve as conversation starters, reflecting the author's ideas and research, and should not be read as Google's official view." On the Institute's own author pages Manyika's title is given as President, Research, Labs, Technology & Society, overseeing Google Research, Google Labs, Learning & Sustainability, and Technology & Society; Legg's as co-founder and Chief AGI Scientist at Google DeepMind; and Hassabis's as co-founder and chair of Google DeepMind and chief scientist of Alphabet (Introducing the DeepMind Institute (Legg, Manyika and Hassabis, September 2026)). No separate governance structure, funding arrangement, staff roster or external advisory board has been described in the launch materials.
Inaugural publications
Five essays were posted at launch (Source: institute.deepmind.com). Authors are as given on each essay page and in The Next Web's launch account (Source: thenextweb.com).
| Essay | Authors | Slug |
|---|---|---|
| Introducing the DeepMind Institute | Shane Legg, James Manyika, Demis Hassabis | introducing-the-deepmind-institute |
| The case for reasoning transparency | Rohin Shah (Director, AGI Safety & Alignment, Google DeepMind); Anca Dragan (VP of AI Safety & Behavior, Google DeepMind) | the-case-for-reasoning-transparency |
| Economic policy for AGI | Julian Jacobs (economist and research scientist, Google DeepMind); Alex Imas (Director of AGI Economics, Google DeepMind) | economic-policy-for-agi |
| Principles for a new utopianism | Stephen Cave (Leverhulme Centre for the Future of Intelligence, University of Cambridge) | principles-for-a-new-utopianism |
| A framework for frontier AI and the dawning of a new age | Demis Hassabis | a-framework-for-frontier-ai-and-the-dawning-of-a-new-age |
Axios described the inaugural publications as covering economic policy for AGI, model reasoning transparency, global access and human flourishing (Source: axios.com). The first two of those descriptions correspond to named essays. Cave's essay describes itself as exploring "how AGI can help establish the conditions for human flourishing" Principles for a new utopianism (Cave, DeepMind Institute, 2026); the mapping of "global access" onto the remaining titles is not stated in any of the sources. Hassabis's framework essay was first published on July 14, 2026 (A Framework for Frontier AI and the Dawning of a New Age (Hassabis, July 2026)).
The essays' arguments, as the pages present them:
- Reasoning transparency. Shah and Dragan argue that chain-of-thought reasoning currently lets developers monitor frontier models for misaligned behavior but "is not guaranteed," citing the GPT-6 Astra system card's reported "substantial decrease in chain-of-thought monitorability." They propose measuring chain-of-thought transparency, preserving transparent architectures, including a possible limit on "opaque serial depth" set by regulation or by developers, and auditing training rewards so the chain of thought is not optimized to look aligned; they propose that developers who deviate from standard chain of thought demonstrate comparable monitorability, and state that regulation could help if commercial goals under-incentivize transparency (The Case for Reasoning Transparency (Shah and Dragan, DeepMind Institute, September 2026)). See Chain-of-Thought Monitorability, Opaque Serial Depth, Reasoning Models and Chain-of-Thought and Monitorability Tax.
- Economic policy. Jacobs and Imas rate 11 policies using literature reviews, surveys and 51 AI agent raters built from survey data on 51 economists, and match three "least-regret" interventions to scenarios: expanded unemployment insurance, Earned Income Tax Credit and employer-led retraining for mild disruption; conversion of the EITC into a negative income tax for moderate displacement; and a universal basic capital backstop if labor's share of GDP declines for a sustained period. They describe universal basic income as "an expensive and blunt instrument," and report survey support among Americans of 85% for publicly funded retraining, 72% for unemployment insurance and 54% for universal basic capital Economic policy for AGI (Jacobs and Imas, DeepMind Institute, 2026). See Universal Basic Income (in AI policy) and AI Labor Disruption.
- Utopianism. Cave argues that "averting apocalypse is not enough: we need visions of utopia too," and proposes three principles for positive visions of a future with AGI — medium-termism, humility and pluralism — offering universal basic services as an example of a project that he suggests may be a better approach than universal basic income Principles for a new utopianism (Cave, DeepMind Institute, 2026).
Statements at launch
In interviews at launch, Manyika framed AGI governance as resistant to national treatment: "Humanity has to be a part of this. I don't know how we think about safety and responsibility on a country-by-country basis. This is a global technology that will be used by everybody." Legg made a parallel argument about scientific progress: "AI is going to drive all kinds of advances in science, and science is an international thing. If some scientific breakthrough happens in China or it happens in Japan or the U.S. or wherever — this is a global community advancing the frontier in science" (Source: axios.com). Legg also told the Financial Times, as reported by The Next Web, that AI capabilities must not outrun safety controls, that it was premature to declare AGI achieved, that he remains "comfortable" with a forecast of a 50% chance of "minimal" AGI by 2028, and that Anthropic chief executive Dario Amodei's call to slow but not pause frontier releases was "interesting directionally" and "worth considering" (Source: thenextweb.com).
The launch fell in a period of active argument over international coordination on frontier AI, including the European Commission's stated intention to invite frontier labs for talks on pacing development and US–China discussions of shared AI risks (Source: reuters.com; axios.com).
Open questions
- The Institute has published an outside author at launch (Stephen Cave of the University of Cambridge), but the editorial terms on which it accepts work from researchers outside Google and Google DeepMind have not been described.
- Whether it has dedicated funding or staff distinct from Google DeepMind's research organization.
- How its output relates to Google DeepMind's existing responsibility and safety publications, including the Frontier Safety Framework.
Relationships
- related: Google DeepMind — parent organization
- related: Shane Legg — director and managing editor
- related: Demis Hassabis — director; author of the republished framework essay
- related: James Manyika — director
- related: Alex Imas — co-author of the inaugural essay on economic policy for AGI
- related: Introducing the DeepMind Institute (Legg, Manyika and Hassabis, September 2026) — founding statement and mission
- related: The Case for Reasoning Transparency (Shah and Dragan, DeepMind Institute, September 2026) — inaugural essay on chain-of-thought monitorability
- related: A Framework for Frontier AI and the Dawning of a New Age (Hassabis, July 2026) — one of the five inaugural publications
- related: Rohin Shah, Anca Dragan — authors of the inaugural essay on reasoning transparency
- related: Artificial General Intelligence (AGI) — the subject the Institute is organized around
- related: Chain-of-Thought Monitorability, Opaque Serial Depth, Reasoning Models and Chain-of-Thought, Universal Basic Income (in AI policy) — subjects of inaugural essays
- related: AGI Timelines — the founding statement expects remaining gaps to AGI "to be closed soon"
- related: The Anthropic Institute — a comparable lab-affiliated publishing body at another frontier developer