Introducing the DeepMind Institute (Legg, Manyika and Hassabis, September 2026)

"Introducing the DeepMind Institute" is the founding statement of the DeepMind Institute (DMI), a publishing forum on artificial general intelligence started by researchers from Google and Google DeepMind. It was published at institute.deepmind.com on September 16, 2026, the day of the Institute's public launch, under the names of its three directors, Shane Legg, James Manyika and Demis Hassabis. The statement runs to about 650 words of body text and is followed by the Institute's mission statement, a listing of its directors and managing editor, a disclaimer, and biographies of the three authors.

Mission

The mission statement says the Institute "aims to advance bold thinking about the safe development of artificial general intelligence (AGI), its beneficial use, and its implications for society." It presents the Institute as working "from the lab that pioneered the field" and as bringing "creative foresight and grounded examination of AI" to what it calls the most important questions of the time. The standfirst states that as AGI nears, interdisciplinary thinking about its implications for humanity is urgently needed.

Argument

Proximity of AGI

The directors write that today's AI systems have impressive capabilities and 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." The phrase "cognitive capabilities" links to an arXiv paper (arXiv:2605.28405). The essay concedes that AI "can still sometimes fail at basic tasks," linking to the SimpleBench benchmark, and that it lacks the consistency and creativity to meet the bar of full AGI, linking both points to a conversation with Hassabis published in Daedalus, the journal of the American Academy of Arts and Sciences. It then states: "we expect those gaps to be closed soon." No date or range is given.

Benefits

The directors state that they have "always believed AGI could prove to be the ultimate tool for accelerating scientific breakthroughs." By opening new paths to understanding and curing disease, developing clean energy and increasing economic prosperity, the essay says, AGI "could unlock a new golden age of scientific discovery and progress" beyond what could be achieved without it.

Risks

The essay names two kinds of risk. The implications of increasingly capable AI for cybersecurity and biorisks are described as already being seen; in addition there is "the potential for loss of control in future self-improving systems." The directors describe the present as the moment to ensure that AGI is built safely and that its benefits to society "far outweigh any risks." See AI and Cybersecurity, AI Biosecurity and Recursive Self-Improvement (RSI).

Purpose of the Institute

The Institute is described as launched "to spur the interdisciplinary research, collaboration and debate required to answer the AGI era's most critical technical and societal questions." The essay poses three:

• what society will value, and how AGI will affect what it means to be human;
• how to build and govern AGI systems safely, together with "the communities of agents they will form";
• which institutions and policies society will need to adapt to AGI, or to reimagine altogether.

DMI is described 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. The directors state that these contributors "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 reach a consensus on the challenges and opportunities society faces.

Who should decide

The essay states that nobody has all the answers about how AGI will be developed and deployed responsibly, and that "it shouldn't be technologists alone who provide them." In its account, technological progress establishes only the possibility of a new era, while "the responsibility of shaping its reality collectively belongs to society as a whole, including the arts and humanities, and governments." The Institute is described as bringing diverse viewpoints together to identify challenges, debate solutions, and help ensure that AGI improves everyone's lives.

Editorial arrangements and disclaimer

The page lists Legg, Manyika and Hassabis as the "DeepMind Institute directors," names Legg as managing editor, and gives an editorial contact address at google.com. Its disclaimer, which also appears on the Institute's other essays including the Shah and Dragan essay on reasoning transparency, 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." The founding statement is therefore a statement by the three directors, all of whom hold senior roles at Google, Google DeepMind or Alphabet, rather than a corporate policy document.

Director biographies

The biographies appended to the essay give the directors' titles as of the launch.

• Shane Legg is listed as co-founder and Chief AGI Scientist at Google DeepMind. The biography records his PhD at the Dalle Molle Institute for Artificial Intelligence Research (IDSIA) in Switzerland, with a thesis proposing a formal definition of machine intelligence that received the Canadian Singularity Institute research prize; a year as a postdoctoral researcher at the Swiss Finance Institute building models of human decision-making; two years at the Gatsby Computational Neuroscience Unit at University College London; an Honorary Fellowship of UCL; Distinguished Alumni status at Waikato University; the 2017 Digital Masters Award for Excellence in Technology; and appointment as Commander of the Order of the British Empire (CBE) in the 2019 Birthday Honours for services to investment and the science and technology sectors.
• James Manyika is listed as President, Research, Labs, Technology & Society, with overall leadership of Google Research, Google Labs, Learning & Sustainability, and Technology & Society. The biography records past service as Vice Chair of the US National AI Advisory Committee, Co-Chair of the UN Secretary-General's AI Advisory Body, Vice Chair of the Global Development Council at the White House, and member of the US Secretary of State's Foreign Affairs Policy Board; fellowship of the American Academy of Arts and Sciences; an inaugural Distinguished Fellowship at Stanford's Human-Centered AI Institute and a Distinguished Fellowship in Ethics in AI at Oxford; and, as a Rhodes Scholar, a DPhil, MSc and MA from Oxford in AI and robotics, mathematics and computer science, with a BSc in electrical engineering from the University of Zimbabwe. Axios's launch account described him as a Google senior vice president (Source: axios.com).
• Demis Hassabis is listed as "Nobel Laureate | Co-Founder & Chair, Google DeepMind | Chief Scientist, Alphabet," the titles he took in the August 2026 leadership change. The biography credits DeepMind's systems under his leadership "as co-founder and CEO," naming AlphaGo and AlphaFold; records the 2024 Nobel Prize in Chemistry for AlphaFold and his founding of Isomorphic Labs; and lists his fellowships of the Royal Society and the Royal Academy of Engineering, his 2024 knighthood, a BA in computer science from Cambridge and a PhD in cognitive neuroscience from UCL.

Relation to other definitions of AGI

The statement's definition of AGI closely follows the one in Hassabis's July 14, 2026 framework essay, which describes AGI as "a system that exhibits all the cognitive capabilities the brain has" and calls it "probably only a few short years away" (A Framework for Frontier AI and the Dawning of a New Age (Hassabis, July 2026)). The founding statement uses nearly the same words but gives no timescale beyond "soon." That essay is one of the Institute's five inaugural publications (Source: thenextweb.com).

The formulation differs from the one in Google DeepMind's June 2026 report From AGI to ASI, on which Legg is a senior author. That report defines AGI as "a system that is roughly as intelligent as a single human… median human-level on most 'cognitive' tasks" and grounds its definitions in the Legg-Hutter measure. "All the cognitive capabilities" is a stricter wording than "most 'cognitive' tasks"; the founding statement does not refer to the report. Gary Marcus argued in June 2026 that Hassabis was running two conflicting timelines, "5 to 10 years" with a strict all-cognitive-capabilities definition at Davos in January 2026 and "2030 plus or minus a year" at Stanford, and read the gap as definitional slippage in which a looser definition yields a nearer date (Source: garymarcus.substack.com). See Artificial General Intelligence (AGI) and AGI Timelines.

Launch coverage

Axios's account of the launch, by Maria Curi, quoted the essay, named Legg as managing editor alongside Manyika and Hassabis, and described the inaugural publications as covering economic policy for AGI, model reasoning transparency, global access and human flourishing. It reported interview remarks from Manyika that safety and responsibility could not be handled on "a country-by-country basis" and from Legg that scientific progress is "an international thing" (Source: axios.com). The Next Web, in an account by Ana Maria Constantin, listed the five inaugural pieces and their authors, and reported Legg's remarks to the Financial Times that it was premature to declare AGI achieved and that he remains "comfortable" with a forecast of a 50% chance of "minimal" AGI by 2028 (Source: thenextweb.com).

Provenance

The text was fetched from the live page on September 17, 2026 and converted from HTML; a Firecrawl copy taken the same day differs in a few standfirst and spelling details, indicating that the page was edited around the launch. The raw file follows the live HTML. The page's article:published_time metadata reads 2026-07-09T15:47:35Z, a content-management date earlier than the public launch, and its article:modified_time reads 2026-09-16T13:49:44Z; the essay is dated here to the launch on September 16, 2026. The document's identity is corroborated by the Axios and Next Web accounts above, and the site is served under Google's privacy and terms policies.

Relationships

• related: DeepMind Institute — the Institute's founding statement and mission
• related: Shane Legg, James Manyika, Demis Hassabis — authors and directors
• related: Google DeepMind — the organization from which the Institute was started
• supports: AGI Timelines — a directors' statement that the remaining gaps to AGI are expected "to be closed soon"
• related: Artificial General Intelligence (AGI) — defines AGI as a system with all the cognitive capabilities of the human brain
• related: A Framework for Frontier AI and the Dawning of a New Age (Hassabis, July 2026) — shares its AGI definition; republished as an inaugural DMI essay
• related: From AGI to ASI (Google DeepMind, June 2026) — Google DeepMind's June 2026 definitions of AGI and ASI
• related: The Case for Reasoning Transparency (Shah and Dragan, DeepMind Institute, September 2026) — inaugural essay published the same day
• related: Recursive Self-Improvement (RSI), AI Biosecurity, AI and Cybersecurity — the risks the statement names