Microsoft AI head's stern warning as tech giants race to build superintelligent system, says 'not going to be a better world if...'

Mustafa Suleyman, the head of AI at Microsoft, has issued a stark warning about the risks of advanced artificial intelligence, saying the world will not be better if humanity loses control of it, adding that raw capability must take a backseat to human control. His remarks came days after the tech giant, who recently laid off its employees, unveiled Microsoft's new MAI Superintelligence Team. He emphasised that while tech firms such as Mark Zuckerberg’s operations are spending billions on AI, raw capability alone is not enough. “It’s not going to be a better world if we lose control of it,” he said.

"It's got to be for humanity's sake, for a future we actually want to live in. It's not going to be a better world if we lose control of it," he was quoted as saying by TOI. Microsoft says it has achieved what Suleyman describes as “AI self-sufficiency”, meaning it is no longer bound by limits on how large a model it could train under its former partnership with OpenAI. Under the old agreement, Microsoft was restricted in model size by computing thresholds measured in FLOPS (floating-point operations per second).

ALSO READ: 'Go work at McDonald's': World's third richest man has an unconventional career tip for entrepreneurs, says 'you don’t need to start...'

'Microsoft taking a different approach'

As companies race to build next-generation systems, Suleyman argued that uncontrolled advancement could be dangerous. He emphasised that Microsoft is taking a different approach — one he calls “Humanist Superintelligence”. He said: “We are not building an ill-defined and ethereal superintelligence; we are building a practical technology explicitly designed only to serve humanity.


Suleyman openly stated that there is no “reassuring answer” yet for how to “contain, let alone align, a system that is—by design—intended to keep getting smarter than us.” He also noted that Microsoft’s cautious path might turn out to be more expensive or less efficient than approaches with fewer safeguards. Mustafa Suleyman indicated it will still be “a good year or two before the superintelligence team is producing frontier models,” emphasising that Microsoft is playing a longer game.

ALSO READ: Comet 3I/ATLAS showed sudden increase in speed and changed colour? New revelations about interstellar object leave scientists surprised

Microsoft’s approach defined

The human-centred approach will focus on AI that is “carefully calibrated, contextualised, within limits” rather than an unrestricted intelligence with high autonomy. Examples include AI for medical diagnostics (where one project reached 85 % accuracy on hard cases versus about 20 % for human doctors), personalised education tools, and clean-energy breakthroughs.