China just put the cofounders of Manus under exit ban while it reviews Meta's $2 billion AI agent acquisition, and this is the real opening shot in the AI sovereignty war.

The Summary

  • Chinese regulators barred Manus cofounders from leaving the country while reviewing Meta's $2 billion acquisition of the AI agent startup
  • This is China deploying its national security review apparatus to control critical AI infrastructure before it leaves the country
  • The exit ban signals Beijing views autonomous agent technology as strategic national assets, not just commercial IP

The Signal

China isn't playing around anymore. The National Development and Reform Commission calling Manus cofounders Xiao Hong and Ji Yichao into a meeting and then barring them from leaving the country is the PRC's way of saying: we decide what AI leaves our borders, not your cap table.

This matters because Manus isn't building chatbots. They're building autonomous agents, the kind that can execute complex workflows without human supervision. That's Web4 infrastructure. And at $2 billion, Meta clearly sees strategic value beyond a talent grab. But China sees something else: a potential national security loss if sophisticated agent architectures and training methodologies flow to American tech giants.

The timing is revealing. We're watching the US tighten export controls on AI chips to China while China responds by weaponizing its regulatory review process to keep AI talent and IP onshore. Exit bans are typically reserved for serious national security cases or major fraud investigations. Using them for M&A review is new territory. It transforms every Chinese AI founder into a potential flight risk the moment they talk to Western acquirers.

This also puts Meta in an impossible position. They've already announced the deal. They've presumably done diligence. But China can drag this review out indefinitely while essentially holding the founders hostage. And if the deal falls apart, those founders now have a target on their backs from both sides: Beijing for trying to sell to the Americans, and the Valley for being un-acquirable.

The Implication

If you're building AI agent infrastructure in China, understand that your exit options just narrowed dramatically. Beijing is drawing a line around autonomous agent technology the same way it did around quantum computing and semiconductor design. For Western acquirers, this is a warning: Chinese AI deals now come with sovereign risk that no amount of diligence can price. The agent economy is fracturing along national lines before it's even fully formed. Watch for more countries to follow China's playbook as they realize letting advanced AI capabilities leave borders is a one-way trip.

Source: The Information