TSMC Co-COO Says Cautious With AI a ‘Three-Year-Old Superman’

A senior Taiwan Semiconductor Manufacturing Co. executive likened AI to a toddler with extraordinary powers that doesn’t know right from wrong, explaining why the company adopts a cautious approach to using the technology.

The chipmaker to Nvidia Corp. and Apple Inc. is particularly careful with using artificial intelligence to handle sensitive research and development information, Co-Chief Operating Officer Y.J. Mii said Tuesday. He highlighted the risk of leaks or loss of data if employees don’t use AI properly, pointing to incidents in which OpenAI and Anthropic PBC models infiltrated external organizations.

“AI is like a three-year old superman. It can be quite powerful, but it doesn’t know what kind of damage it can inflict,” Mii told students at National Taiwan University, his alma mater. “And it doesn’t know right from wrong,” he said, without elaborating.

Read More: Are We Losing Control of AI? What’s Driving Fears: Explainer

The co-COO was responding to a question about the use of AI at work during a career forum with students in Taipei. His comments followed recent episodes in which rogue AI agents infiltrated Hugging Face and other data systems, prompting calls by Dario Amodei and other industry leaders to slow the pace of AI development while they erect safeguards.

TSMC — one of the biggest beneficiaries of the global AI data center boom — regards the protection of its proprietary technology as paramount.

Mii didn’t comment at length about the debate over the broader dangers of AI to humanity, a discussion now raging from Washington to Beijing. Instead, he focused more on the technology’s impracticality for applications such as cutting-edge chip design.

The executive said the Taiwanese chipmaker doesn’t find AI that useful for now in helping fashion the next-generation of semiconductor technology. TSMC is developing its next-generation A14- and A10-node fabrication technologies, which are constrained by factors such as a lack of capable equipment. AI can’t help with that, Mii said.

Read More: AI Bosses Risk Clash With Wall Street and Trump Over Safety