China reviews Meta’s $2bn AI deal over export control concerns

Summary:

• China reviewing Meta’s $2bn acquisition of AI firm Manus

• Focus on whether technology transfer required export licence

• Review is preliminary, not yet a formal investigation

• Manus gained attention for autonomous AI agent technology

• Case highlights rising geopolitical risk around AI deals

Chinese authorities are reviewing Meta’s roughly $2 billion acquisition of AI start-up Manus for potential violations of China’s technology export control rules, raising fresh uncertainty around one of the year’s most high-profile artificial intelligence deals.

According to people familiar with the matter, officials at China’s Ministry of Commerce of China have begun assessing whether the relocation of Manus’s staff and core technology to Singapore, ahead of its sale to Meta, should have required an export licence under Chinese law. While the review is still at an early stage and may not evolve into a formal investigation, the licence question alone could give Beijing leverage over the transaction.

Manus, which is now based in Singapore, attracted widespread attention earlier this year after releasing what it described as the world’s first general AI agent, software capable of autonomously making decisions and executing tasks with minimal prompting, setting it apart from traditional chatbots. The product’s viral reception underscored growing interest in agent-based AI systems and their potential commercial and strategic value.

Meta completed the acquisition last month, with sources indicating the deal valued Manus at between $2 billion and $3 billion. Neither Meta nor Manus has commented publicly on the regulatory review, and Reuters said it could not independently verify the Financial Times (gated) report.

For Beijing, the case highlights increasing sensitivity around advanced AI capabilities and the movement of high-value technology overseas. China has steadily tightened its export control framework in recent years, particularly for technologies deemed strategically important, including advanced semiconductors, AI models and data-intensive systems.

While officials have not accused the companies of wrongdoing, analysts note that an export-licence requirement could, in an extreme scenario, allow regulators to delay, condition or even pressure parties to unwind the transaction. More broadly, the review underscores the growing regulatory and geopolitical risks facing cross-border AI deals, especially those involving Chinese-origin technology and major U.S. tech firms.