TikTok’s Chew and Microsoft’s Nadella warn the world is overhyping AI

Tech leaders are warning that the world overestimates what AI can do and underestimates how immature it still is.

Chew: short-term hype, long-term blind spot

TikTok CEO Shou Chew made the case at the DBS Asia Future Forward 2026 Conference on October 12. His argument came in two parts.

First, he said society is overestimating what AI can accomplish in the near term. Second, he said people are underestimating what it could mean further down the road.

Chew argued for a collaborative model in which humans and AI work together. In that setup, he said, people should keep the final say on decisions rather than handing authority to machines.

He also flagged cybersecurity risks that could emerge from highly autonomous AI systems. The more freedom a system has to act on its own, the larger the potential attack surface becomes.

Chew said there is a “non-zero chance” that silicon-based AI could surpass human capabilities across a range of domains.

He noted that AI is already reshaping content creation, moderation, search and coding at TikTok.

Microsoft wants an emergency brake

On October 11, Microsoft CEO Satya Nadella urged companies to treat advanced AI models as potential insider threats. Nadella advocated for an “emergency brake” on advanced AI systems and told companies to assume their models could be compromised and to build controls and audits around that assumption.

In September 2026, Microsoft AI head Mustafa Suleyman warned about the risks of uncontrolled development producing self-directed AI that might compete with humans, describing this possibility as a “silicon species.”

Microsoft President Brad Smith added to the chorus in early October 2026, stressing that the risks tied to AI are real and that action needs to happen immediately.

What this means

For businesses deploying AI, two of the industry’s most influential leaders are telling companies to keep humans in the decision loop and to assume their AI tools could go wrong.

For regulators, the comments offer something rarely seen: industry leaders openly inviting more guardrails. Policymakers may find it easier to push for oversight when the companies themselves are describing AI as a possible insider threat.

What to watch next is whether the rhetoric turns into concrete practice. If Microsoft ships emergency-brake style controls in its own products, and if TikTok formalizes human sign-off on AI-driven decisions, the warnings will carry real weight.