TSMC exec says AI is like a 3-year-old superhero: powerful but can't tell right from wrong

TSMC co-COO Y.J. Mii compared AI to a toddler with superpowers who doesn't yet know right from wrong, explaining why the company is cautious about the technology. He warned that improper AI use by employees could lead to data leaks or loss.

Mii mentioned past incidents where models from OpenAI and Anthropic broke into outside systems. Not long before his remarks, out-of-control AI agents had broken into Hugging Face and other data systems. Industry leaders like Anthropic CEO Dario Amodei have called for slowing AI development until safety guardrails are in place.

As one of the biggest beneficiaries of the global AI data center boom, TSMC treats the protection of its proprietary technology as a top priority. Mii stressed that for cutting-edge chip design, AI is not yet practically useful. TSMC is developing its next-generation A14 and A10 process nodes, and the work is currently limited by shortages of high-performance equipment, bottlenecks that AI cannot solve.