Beijing is seeking to clarify its decision to block Meta’s proposed acquisition of Manus, a Chinese agentic artificial intelligence (AI) startup, stressing that it will continue to support domestic companies’ overseas expansion, provided the expansion is not structured as so-called Singapore washing.

State media have published a series of commentaries to explain the policy rationale after the National Development and Reform Commission’s Office of the Working Mechanism for Security Review of Foreign Investment on Monday formally prohibited the foreign takeover of Manus in accordance with laws and regulations, ordering the parties to unwind the transaction.

Chinese commentators said Beijing does not want its decision on Manus’ case to send wrong signals to foreign investors that China is tightening control over inbound investment or restricting domestic firms’ overseas expansion.

In an article published on Thursday, Yuyuan Tantian, a social media account affiliated with China Central Television, said China’s Measures for the Security Review of Foreign Investment clearly define the scope of scrutiny.

“Under Article 4, investments involving national defense security must be declared regardless of foreign ownership levels, while in key sectors such as important information technology, internet products and services, and critical technologies, any foreign investor gaining actual control falls within the review scope,” it said.

It added that Manus’s general-purpose AI agent falls squarely within these key technology categories, and that a takeover by Meta would have resulted in actual control, and that such transactions are required by law to be proactively declared by the companies involved. It said Manus did not declare the transaction to the Chinese authorities.

Citing legal and technology experts, the article said regulators would primarily assess three layers of risk: technology, talent and data.

“Manus’s core assets, including algorithms, data and talent, were developed within China by domestic teams, and any transfer of control overseas would require a national security review,” it said.

“At present, some countries are expanding the scope of security reviews and blurring the definition of threats, specifically targeting the AI development of other countries. This value orientation is influencing their narrative of security,” the article said, without naming the United States, which has banned American funds from investing in Chinese AI firms and imposed chip export controls on China.

“In order to safeguard their own security, they may even use other countries’ capabilities to attack them. We have to be vigilant,” it stressed. “Meanwhile, China will continue to encourage AI innovation and remain open to foreign investment.” 

In the past, Chinese tech firms often pursued dual listings, in Hong Kong and on NASDAQ, to fund expansion. Since October 2024, however, US restrictions have barred American funds from investing in China’s AI sector, giving rise to the workaround called “Singapore washing.” The term refers to spinning off or relocating to Singapore to raise capital.

In Manus’s case, the company severed ties with China to secure Meta’s investment, but that approach has now proven ineffective.

Three red lines

Manus first drew global attention with its high-profile debut in March 2025, surprising the global technology sector by demonstrating its ability to complete tasks traditionally performed by white-collar workers. Developed by Beijing-based startup Butterfly Effect, the system was positioned as a general-purpose AI agent rather than as a conventional large language model (LLM) such as ChatGPT or DeepSeek.

In promotional footage, Butterfly Effect co-founder Xiao Hong demonstrated the system’s practical capabilities, showing how the AI agent could screen 10 resumes, find a New York property within a set budget, and analyze correlations among shares of Nvidia, Marvell Technology and Taiwan Semiconductor Manufacturing Co (TSMC), demonstrating its ability to handle complex, multi-step tasks. The company’s slogan was: “Leave it to Manus.”

However, in late March 2026, Xiao and co-founder Ji Yichao were reportedly barred from leaving China following a meeting with the National Development and Reform Commission (NDRC) in Beijing, as the Ministry of Commerce initiated a national security review of Meta’s proposed US$2 billion acquisition. On April 27, China formally banned Meta’s acquisition of Manus. 

Chinese media subsequently outlined the company’s restructuring over the past year:

• In June–July 2025, Manus moved its headquarters to Singapore and switched its operating entity to Butterfly Effect Pte. It cut its mainland team from more than 120 staff to about 40 core members, who were relocated to Singapore, while clearing its China-based social media accounts and blocking China’s IP addresses from accessing its website.
• By late 2025, it appeared to operate as a Singapore-based company. On December 30, Meta announced a US$2 billion acquisition, with Xiao Hong expected to take a senior role in the US company. 
• In January 2026, regulators intervened, launching a review that ultimately led to the deal being blocked.

A Guangdong-based business columnist writing under the pen name “Shengchandui” says Manus has crossed three key red lines in China. He lists the three as technology sovereignty, data sovereignty and national security.

“Where the technology originates determines jurisdiction,” he says, “Manus’s core algorithms and team were built in China. Shifting the company offshore and selling it to a US buyer amounts to exporting domestically developed capabilities. It is a form of ‘technology smuggling’ that could undermine China’s own innovation base.”

The columnist continues: “Data sovereignty cannot be compromised. Manus processes vast amounts of data, much of it from Chinese users. Transferring control overseas risks turning technology outflow into potential data leakage, especially under stricter rules governing cross-border data transfers.” 

He adds further that allowing a system built on Chinese technology and data to come under foreign control could pose significant risks to China’s national security, as AI agents are evolving into core components of digital infrastructure across work, communication and software development.

“This is not a ban on Chinese firms’ global expansion plans, but a ban on evading regulation,” said Zhu Youping, a researcher at NDRC’s State Information Center. “If the proposed acquisition is completed, Meta would obtain 100% control in Manus, but neither Meta nor Manus had declared this to the Chinese regulators,” he said.

“Regulators are applying a ‘look-through’ approach, focusing on the origin of technology, the source of data and the ownership of talent, rather than the company’s place of registration,” Zhu said. “Manus’s relocation to Singapore is essentially a case of using domestic resources to incubate value and monetizing it through an offshore structure to bypass oversight.”

AI agents for service sectors

Beyond preventing foreign control of Manus, Beijing also wants the company to remain in China and contribute to the development of the domestic AI industry.

“China’s AI industry has entered a phase of rapid development, with a sustained burst of innovative vitality, making it a fertile ground for global AI innovation,” the Global Times said in its editorial on Tuesday. “We hope that more technology and innovation enterprises, including Manus, can find their place in this blue ocean in China, develop confidently, grow larger and stronger and achieve better development and breakthroughs.”

On April 21, the State Council issued a policy document outlining 20 measures to expand and upgrade the sector, setting a target for total output to exceed 100 trillion yuan (about US$13.8 trillion) by 2030 from 81 trillion yuan last year. It means the country will need to add about 4 trillion yuan to its annual output.

The policy specifically supports the use of AI tools in areas such as intelligent programming, contract review, financial services and supply chain optimization, while also calling for the development of national AI application testing bases.

Pang Chaoran, a researcher at the Chinese Academy of International Trade and Economic Cooperation (CAITEC), said the shift marks a clear change in policy direction, from subsidizing companies to train AI models to encourage private firms in the service sectors to use AI models and agents.   

He said, by encouraging businesses to adopt AI tools at scale, Beijing aims to accelerate commercialization and embed AI more deeply into real economic activity, creating new growth momentum for both the service and technology sectors.

Read: After DeepSeek: China’s Manus – the hot new AI under the spotlight

Follow Jeff Pao on X at @jeffpao3