Who is Mustafa Suleyman?
Mustafa Suleyman is a British technology entrepreneur and CEO of Microsoft AI. He cofounded DeepMind and Inflection and has led applied research, product development and technology commercialisation. His career connects the creation of AI laboratories with public debate about controlling increasingly capable systems. [4]
Education and early social ventures
Born in August 1984, he is recorded in the British company register as a Muslim Youth Helpline director from 2004 to 2008. He subsequently belonged to Reos Partners LLP from 2008 to 2010 and directed Reos Partners CIC from 2009 to 2010. These social and consultancy roles preceded DeepMind. [5]
He entered Mansfield College, Oxford, in 2002. The university later included him among alumni honoured for contributions to British technology. [6]
He left Oxford at 19 to help establish a nonprofit telephone counselling service, without completing his university degree. [3]
DeepMind: from founding to applied research
In 2010 he cofounded DeepMind with Demis Hassabis and Shane Legg, bringing different scientific disciplines together to investigate more general learning systems. [7]
In the 2016 review he signed with his cofounders, he described advances including AlphaGo, WaveNet speech synthesis and machine learning for data centres. These collective projects illustrate the laboratory’s breadth, spanning games and perception as well as industrial operations. [8]
Data-centre cooling evolved from recommendations for operators to autonomous control in 2018. The system checked proposed actions against local limits and allowed operators to retake control. The development belongs to DeepMind’s technical team; Suleyman is not credited with individually inventing its algorithm. [9]
DeepMind Health, participation and clinical data
With Dominic King, he promoted patient participation in DeepMind Health. In September 2016 they presented a forum for patients and the public to discuss priorities, trust and design. Their approach sought to involve people receiving care alongside professionals using the tools. [10]
In 2017 they acknowledged shortcomings in communication and engagement following the ICO’s investigation into Royal Free data used to test Streams. They explained that they had underestimated the health system’s complexity and the need for accountability to patients. The investigation concerned the hospital’s processing of data. [11]
In 2019 they presented research using historical Veterans Affairs data to predict acute kidney injury up to 48 hours ahead. This was a retrospective model evaluation, distinct from the Streams clinical-alert application. The study and service should not be conflated into a single care intervention. [12]
The Streams team’s integration into Google Health was announced in November 2018 and completed in September 2019. It aimed to draw on Google’s infrastructure and experience to expand the tool. Hassabis, Suleyman and King signed the announcement. [13]
Google, Inflection and the Pi assistant
In 2019 he moved to Google as a vice president responsible for AI products and policy. His Harvard Belfer Center profile records this period and his Senior Fellow position, associated with studying emerging technologies’ public consequences. [14]
In 2022 he founded Inflection AI with Reid Hoffman and Karén Simonyan. The company introduced Pi on 2 May 2023 as a personal AI for conversation, learning and exploring ideas. Suleyman was its CEO; the product combined proprietary models with an interface designed for everyday dialogue. [15]
Inflection argued that an assistant should support a person’s intentions and goals instead of optimising attention alone. Its introduction contrasted this aspiration with attention-driven digital business models. The document describes a product philosophy, rather than an independent measurement of its effects on users. [16]
The company incorporated as a Public Benefit Corporation. Explaining this model, it linked commercial objectives to public benefit and consideration of affected communities. This corporate form established a governance framework; it did not itself demonstrate that every use of Pi would benefit its users. [17]
Microsoft AI and changing responsibilities
Microsoft hired him in March 2024 as executive vice president and CEO of Microsoft AI, initially responsible for Copilot, Bing and Edge. Simonyan became chief scientist and other Inflection employees joined the team. Suleyman reported to Satya Nadella. [2]
Britain’s competition authority examined the hiring and associated arrangements between Microsoft and Inflection. On 4 September 2024 it cleared the transaction without referring it to a second-phase investigation. The case concerned these specific arrangements, rather than an acquisition of the entire Inflection company. [18]
In March 2026 he shifted focus to models and superintelligence as Jacob Andreou took unified responsibility for Copilot. Suleyman remained Microsoft AI’s CEO. [1]
In-house models and superintelligence research
In August 2025 Microsoft AI announced MAI-Voice-1 and MAI-1-preview. The first was used in Copilot voice experiences; the second entered public testing. These releases expanded in-house model development alongside Microsoft’s partnerships with other AI providers. [19]
In November 2025 Suleyman introduced a team devoted to what he called humanist superintelligence. He advocated domain-focused systems designed to benefit people while remaining under human control. This was a research and product direction, rather than a claim that general superintelligence had already been achieved. [20]
In June 2026 he announced seven new MAI models covering images, voice, transcription, coding and reasoning. His explanation emphasised iteration between training, evaluation and use, accompanied by technical and safety reports. The family brought together different capabilities rather than reducing them to one performance score. [21]
MAI’s draft code of conduct, released for consultation in September 2026, proposed that models accept human correction and interruption and refrain from independently expanding their goals. It defined intended behaviours and development criteria; establishing whether systems meet them requires evaluation of particular models. [22]
The Coming Wave and technology debate
In 2023 he published The Coming Wave with Michael Bhaskar. It examines the spread of AI and synthetic biology and the challenge of containing powerful technologies without losing their benefits. [3]
Honours and institutional roles
He received a Commander of the Order of the British Empire honour in 2019 for services to the UK technology industry. [6]
He has been a non-executive director of The Economist Group since June 2019 and serves on its technology investment committee. This oversight role complements his executive work in AI and public contributions on technology’s economic and social consequences. [23]
