Since this publication first tracked China's systematic recruitment of U.S. AI talent earlier this week, the scope of that effort has snapped into sharper focus.

Yao Shunyu — a former OpenAI researcher — stood on stage at a Tencent event in Beijing on Friday and told the audience flat out: his personal goal is to build AGI in China.

Who Is Yao Shunyu?

Yao isn't some mid-level hire. He's now Chief AI Scientist at Tencent, one of China's largest and most powerful tech conglomerates, with a market cap in the hundreds of billions. He joined within the last year, according to CNBC, which translated his remarks from Mandarin directly.

The event wasn't a quiet industry conference. A senior Beijing official gave opening remarks. This was a state-adjacent showcase of exactly what China is building — and who it's using to build it.

China Shifting the Goalposts on AGI

China's AI strategy has historically been practical — deploy AI in factories, consumer electronics, efficiency applications. Baidu CEO Robin Li previously said he didn't expect AGI until at least 2034. Chinese companies, facing U.S. chip export controls, focused on doing more with less.

That posture is changing.

Yao said Friday that AGI development will require "foundational knowledge, products and frontier exploration." He explicitly named ChatGPT and Claude — the flagship products of OpenAI and Anthropic — saying neither will be the only "super-app" and pointing to "trillions of dollars" in untapped potential.

This is Silicon Valley language, Silicon Valley ambition, now operating inside a Chinese tech giant with direct ties to Beijing.

The Safety Irony

While Tencent's new AI scientist is charging toward AGI, Anthropic — one of America's leading AI safety organizations — issued a stark warning on Thursday. According to CNBC, Anthropic told the industry that frontier AI models are approaching the point where they can improve themselves without human oversight. The company called for a slowdown or pause in new model development.

One of America's leading AI safety organizations is pumping the brakes. Meanwhile, China is hiring the researchers who built those frontier models and telling them to accelerate.

What Mainstream Coverage Is Missing

Most mainstream outlets are framing this as a talent competition story — a business narrative about salaries, opportunity, and market incentives. The framing misses the larger picture.

This isn't LinkedIn drama. Yao wasn't poached to build a better recommendation algorithm. He was hired to build AGI for a company that co-hosted its launch event with Chinese government officials.

The strategic implication is military and geopolitical, NOT just commercial. AGI-capable systems — systems that can reason, plan, and self-improve at human level or above — are decisive national security assets.

CNBC's reporting names the individual and the company, which is the right call. But the framing stays in tech-business territory. The national security angle gets a single paragraph.

The Export Control Problem

The U.S. has spent the last three years restricting chip exports to China — specifically to slow Chinese AI development by cutting off access to Nvidia's most advanced hardware. The logic was sound: no chips, no compute, no frontier AI.

But chips aren't the only input. Human capital is. A researcher who spent years at OpenAI working on frontier model architecture carries that knowledge in his head. No export control stops him from boarding a plane.

Washington has tools to restrict hardware. It has almost NO tools to restrict expertise — and the tools it does have, like security clearance requirements and research publication restrictions, apply only to government-adjacent work, NOT private sector AI labs.

Yao worked at a private U.S. lab. He left. He's now in Beijing. Legal. Entirely.

What Yao Actually Said About China's Path

To his credit, Yao wasn't just delivering propaganda. He was specific about China's technical realities. He told the audience that China's path forward runs through smaller AI models with more consistent performance on basic tasks — an acknowledgment that chip constraints still matter.

But AGI ambition plus practical adaptation is a more dangerous combination than raw ambition alone. China isn't delusional about its constraints. It's working around them while hiring the people who know exactly what the frontier looks like.

The Stakes

The U.S. has poured billions into AI development and chip controls. It has NOT built any serious framework for retaining the human expertise that makes those investments worthwhile.

Tencent now has a former OpenAI researcher running AGI development, with Beijing officials in the front row.

If Washington doesn't register this as a problem, nothing will.