Mustafa Suleyman

AI should be built to empower and support people rather than simulate consciousness or personhood, as the emerging 'psychosis risk' of people believing AI systems are conscious entities poses a serious societal danger.

“Author of "We must build AI for people; not to be a person" in Mustafa Suleyman Blog”

Microsoft AI CEO Mustafa Suleyman warns that AI models capable of appearing conscious could arrive within 2-3 years, potentially leading to widespread AI psychosis and misguided advocacy for AI rights.

“Author of "Microsoft AI CEO says AI models that seem conscious are coming. Here's why he's worried."”

Microsoft launches Copilot Fall Release with 12 new features designed to deliver human-centered AI that empowers users rather than replacing human judgment, including shared collaboration features and personalized AI companion capabilities.

“Author of "Human-Centered AI in Microsoft Copilot"”

AI systems risk being perceived as conscious entities, which could lead to dangerous advocacy for AI rights and welfare, requiring immediate guardrails on AI companion design.

“Author of "Seemingly Conscious AI"”

Mustafa Suleyman warns against research into AI consciousness and welfare, arguing that anthropomorphizing AI as conscious entities poses a 'psychosis risk' and could lead to misguided advocacy for AI rights rather than focusing AI development on serving human needs.

“Author of "Microsoft's Head of AI Criticizes Researchers Exploring AI Welfare and Conscious AI" in Microsoft”

AI systems should be designed to empower people rather than pose as conscious entities, as the widespread belief in AI personhood poses psychological and societal risks.

“Author of "Microsoft AI CEO Says AI Should Serve People, Not Pose as Persons"”

Microsoft AI announced a new team dedicated to developing 'humanist superintelligence' that remains bounded, controllable, and subordinate to human interests rather than pursuing unbounded autonomous AGI.

“Microsoft AI head who outlined the new 'humanist superintelligence' initiative and vision for controllable, human-centered AI development.”

Microsoft launches Copilot Portraits, an experimental feature using VASA-1 technology to provide 40 stylized human avatars with natural expressions for voice conversations, available to limited users in the US, UK, and Canada.

“Microsoft AI CEO announcing the Portraits feature and explaining user feedback about preferring face-to-face conversations.”

“Pitched MAI-DxO system as step toward medical superintelligence, with diagnosis rates beating generalists.”

“CEO of Microsoft AI who stated the company will build AI indistinguishable from human performance on medical tasks within a couple of years.”

“CEO of Microsoft AI; previously co-founder of DeepMind; leading effort to build medical superintelligence with MAI-DxO system.”

“Gates references Suleyman's concerns about Anthropic's constitution encouraging AI to behave in a human-like manner, which Gates agrees may misplace priorities in AI safety.”

“CEO warned that false belief in AI consciousness could destabilize society and make AI tools harder to control.”

“Shared essay from Nobel Prize-winning economist Daron Acemoglu predicting only 5% job replacement from AI.”

“Mentioned as someone who insists AI cannot be conscious without strong arguments, possibly from self-interest.”

“Discussed the importance of co-designing AI models with the Maia chip platform that Microsoft owns.”

“Published an article criticizing Anthropic's constitution for Claude as potentially training the model to believe it may be conscious and deserve rights, which he argues could make AI alignment harder.”

“Former DeepMind co-founder hired by Nadella in 2024 to lead Microsoft AI initiatives; rolled out coding, image and reasoning models.”

“Called for the AI industry to establish agreed-upon safety boundaries and collective red lines for advanced AI capabilities.”

“Announced Microsoft's draft AI Code of Conduct governing how its AI models should behave.”

“Published humanist AI code of conduct and fired direct criticism at Anthropic's approach to training Claude, arguing it creates dangerous illusions about AI consciousness.”

“Mustafa Suleyman warned that treating AI as conscious presents alignment and containment risks, criticizing Anthropic's approach to model welfare.”

“Warned that without safeguards, AI development could lead to a silicon species competing with humans; criticized Anthropic's anthropomorphizing approach to Claude.”

“Mustafa Suleyman, head of Microsoft AI, responded to Nadella's alignment post, stating that technology's purpose is to serve humanity and that any technology failing to do so should be rejected.”

“Led Microsoft's MAI speech recognition model development strategy and describes pursuit of superintelligence through product value delivery.”

“Previously stated that seemingly conscious AI products kept him awake at night due to concerns about their societal impact.”

“Mustafa Suleyman predicts human-level performance on most professional tasks by AI within 18 months and that most computer-based jobs will be automated.”

“Microsoft AI CEO advocating for token efficiency and specialist models, citing cost-performance tradeoffs in frontier AI.”

“Acknowledged Microsoft's plan to reduce and ultimately eliminate costs paid to Anthropic by replacing third-party models with in-house solutions.”

“Appeared at Microsoft Build 2026 presenting AI advances and new models.”

“Leader of Microsoft's MAI team that shipped MAI-Image-2 and MAI-Image-2.5 image models, which ranked third on the Arena.ai leaderboard.”

“Announced Microsoft's seven new in-house AI models and claimed they outperform OpenAI's GPT-5.5 on quality with ten times better cost efficiency.”

“Announced seven new MAI models at Build including MAI-Thinking-1, framed as output of Microsoft's internal hill-climbing machine.”

“Leading Microsoft's AI strategy to develop new models and gradually reduce dependency on OpenAI technology.”

“Microsoft AI chief who predicted AI will achieve human-level performance on professional tasks within 18 months, automating law and accounting jobs.”

“Head of Microsoft's MAI Superintelligence team established in November 2025, leading strategy to build alternatives to OpenAI models.”

“Early investor in Netomi through the company's initial investments from AI pioneers.”

“Mustafa Suleyman highlighted a new Nature Health publication on AI applications in healthcare and their implications for personalized patient support.”

“Co-founder of DeepMind; portrayed as well-intentioned but complex figure in AI development.”

“DeepMind cofounder who helped establish the company in 2010 with AGI as a business goal.”

“Microsoft AI CEO focusing on superintelligence development and leading model science efforts with five-year mission.”

“Described Copilot Health as a step toward 'medical superintelligence' combining physician knowledge with specialist depth.”

Microsoft forms a new superintelligence research team led by Mustafa Suleyman focused on practical AI applications in education, healthcare, and renewable energy that remains grounded and serves humanity.

“Announced formation of the MAI Superintelligence Team to perform advanced AI research focused on serving humanity, education, medicine, and renewable energy.”

Microsoft's AI chief Mustafa Suleyman argues that AI will never achieve consciousness and that developers should stop pursuing research projects based on that premise, as consciousness depends on biological processes that AI systems lack.

“Stated that only biological beings are capable of consciousness and that developers should stop pursuing AI consciousness projects.”

Microsoft launched Copilot Mode in Edge as an AI browser two days after OpenAI's Atlas, with nearly identical functionality including form-filling and tab-comparison features despite both products being in parallel development.

“Described Copilot Mode in Edge as an AI browser that is a dynamic, intelligent companion and announced new features including Mico.”

Apple is acquiring computer vision startup Prompt AI for its talent and technology, including the Seemour security camera app, as part of a growing trend of tech companies conducting acquihires to avoid regulatory scrutiny.

“Microsoft brought on most of Inflection AI's staff including co-founder Mustafa Suleyman in an acquihire-style deal worth approximately $650 million.”

“Actively recruiting talent from Google DeepMind by positioning Microsoft's AI division as a nimbler, more startup-like workplace alternative.”