Microsoft warns AI could become a rival ‘silicon species’

What happens if an AI system can set its own goals, earn its own money, and own its own assets? According to Microsoft's AI chief, you'd essentially be creating a new species, and one that might not have humanity's best interests at heart.

In the interview, Mustafa Suleyman went after Anthropic directly, accusing the company of taking the wrong path in its approach to AI, which entails creating technology that shares the human traits. Suleyman is not only against it philosophically; he is concerned that giving artificial intelligence the characteristics of humanity will make it even more difficult to control it.

In an essay written and published earlier this week, Mustafa Suleyman was clear about what artificial intelligence actually is. He said that AI consists of systems that do not have any consciousness, emotions or internal motivations.

Instead, it is nothing but a machine designed to perform the tasks assigned to it by humans.

Suleyman is pushing for AI that stays firmly "subordinate" to human control, backed by more transparency in how models are trained and evaluated, plus independent scrutiny and stronger monitoring tools.

That vision lines up with Microsoft's newly published Humanist AI Code of Conduct, which frames its goal as building advanced AI that stays within limits and remains under human control.

Dame Wendy Hall of the University of Southampton welcomed the substance of the debate but pushed back on alarmist framing from parts of the AI industry, arguing it risks scaring the public rather than informing it.

It's a reminder that even AI leaders who agree on the destination, safe, controllable systems, disagree sharply on how loudly to sound the alarm along the way.