Microsoft AI CEO: Let's control AI before trying to make it "good"

Technology Jan 11, 2026

Mustafa Suleyman, the head of Microsoft AI, says we need to focus on "containment"—basically putting limits on what AI can do—before worrying about making sure it shares our values.
In a recent post on X, he warned that "You can't steer something you can't control," emphasizing concerns about the rapid pace toward superintelligent AI.

Shifting the focus: Practical uses and human oversight

Instead of chasing wild superintelligence, Suleyman wants to use advanced AI for things we can actually manage—like improving medical diagnostics or clean energy.
He points out that Microsoft's AI already hits 85% accuracy on tough medical cases.
Plus, he's building a team (thanks to a new deal with OpenAI) dedicated to keeping humans in charge as this tech grows.