Skip to content

AI Governance

China just forced Meta to unwind a $2 billion AI acquisition, and Manus users have a real data deadline. Here is what business owners should know before trusting any AI vendor's ownership to stay the same.

On August 11, 2026, Manus, an AI agent platform that autonomously browses the web, researches, and writes reports, told users it will soon resume operating as an independent company. The announcement completes the unwinding of Meta's roughly $2 billion acquisition of the startup, after China's National Development and Reform Commission ordered in April 2026 that the deal be withdrawn on national security and foreign-investment grounds. The direct answer for a business owner: some Manus users now have a real, dated deadline. Data generated on or after December 29, 2025 will be deleted for certain users, and those users have until August 23, 2026 to back it up. This did not happen because Manus was hacked or because the product failed. It happened because a government ordered a corporate ownership change reversed, and the company's actual customers are the ones now scrambling to save their work.

By Fabio Rabelo · Founder, ATLACIS ·

What happened

In December 2025, Meta announced it had agreed to acquire Manus, an AI agent platform that launched in March 2025 and lets users hand off tasks like web research and report writing to an AI system that works largely on its own. Manus was built by Butterfly Effect, a company founded in Beijing in 2022 that relocated its headquarters to Singapore in mid-2025, a move that involved laying off China-based staff and reducing its Chinese online presence. The deal drew scrutiny almost immediately. In April 2026, China's National Development and Reform Commission, the government body that oversees economic planning and plays a central role in the country's AI policy, ordered the transaction withdrawn, citing rules on foreign investment and national security. Meta said at the time that the transaction had complied fully with applicable law, and the two companies began a long unwinding process. Bloomberg reported in June 2026 that Meta had already cut Manus staff off from its internal systems and barred its own employees from using Manus tools, an operational firewall ahead of a full separation. On August 11, 2026, Manus told users it will soon resume operating as an independent company. In the same notice, the company said that data generated by certain users on or after December 29, 2025, the date the acquisition was announced, will be deleted, to comply with regulatory requirements in specific parts of the world. Affected users have until August 23, 2026 to back up their data. Users not affected can keep using the service without disruption. Reuters has separately reported that Chinese internet company Tencent is in talks to become Manus' largest shareholder as it returns to independent operation.

Why it matters for business owners

Manus is not an obscure product. It is a widely used AI agent platform that went viral for its ability to complete multi-step tasks with minimal supervision, and businesses that adopted it did so the same way they adopt any AI tool: because it was useful, not because they had mapped who ultimately owned it or which country's regulators could order its ownership reversed. That is exactly the gap this story exposes. A real product, with real business users, is now deleting real customer data on a real deadline, not because of a security failure or a policy the business agreed to when it signed up, but because a national regulator in a country the business may have no connection to ordered a corporate transaction unwound. Any business that had workflows, research, or reports sitting in Manus and did not see this coming is finding out the hard way that vendor risk is not only about pricing and uptime.

What owners should not misunderstand

This is not a data breach. Manus was not hacked, and there is no indication any customer data was accessed by an unauthorized party. The deletion is a company complying with a regulatory order tied to a corporate ownership dispute, not a security incident. This is not evidence that Manus, or AI tools with origins outside the United States generally, are unsafe or lower quality. The dispute here is about foreign investment rules and national security review of a specific acquisition structure, not about the underlying technology or how it handles data day to day. This is not a story only about companies using Chinese-founded AI tools. The mechanism that matters here is broader: a government blocked a large tech company's acquisition of an AI startup and forced a real, disruptive unwinding more than a year after the deal was announced. Any AI vendor whose ownership, funding, or corporate structure crosses a border a regulator cares about carries a version of this same exposure, regardless of which country is involved.

The operational lesson

Most businesses evaluate an AI vendor on what the tool does, what it costs, and how well it fits the workflow. Almost none evaluate who owns the company behind it, whether that ownership is contested, or what happens to their data if a regulator forces a change nobody at the business agreed to or was warned about in advance. Manus users found out about an active data deletion deadline through a company notice, after the ownership dispute had already been working through two governments for months. The unwinding process itself took over a year from announcement to completion. A business relying on any AI vendor whose ownership sits inside a live regulatory dispute, in any country, is exposed to that same kind of disruption, usually with far less notice than the roughly twelve days Manus gave affected users here.

What a serious business should do next

List the AI vendors your business actually depends on for real work, not just the ones with a subscription line item, and for each one, find out who owns the company today. A quick search for recent ownership or acquisition news on each vendor takes minutes and surfaces exactly this kind of exposure before it becomes an active deadline. Build a habit of exporting or backing up anything important that lives inside an AI tool, research, reports, generated work product, on a regular schedule, independent of any specific vendor news. The businesses least affected by an event like this are the ones that already treat their AI tools' output as something to keep a copy of, not as a permanent home for their work. Before adopting a new AI tool for a workflow that matters, add one question to the usual price-and-features evaluation: is this company's ownership or funding structure involved in any active regulatory dispute, in any country, that could force a change to how the product operates or what happens to customer data. If the answer is unclear, that uncertainty itself is worth weighing against how central the tool would become to the business.

The Atlacis view

This is not a reason to avoid a category of AI tools based on where a company was founded. It is a reminder that a vendor's corporate ownership and jurisdiction are part of the same risk picture as its pricing, its data handling, and its uptime, and most businesses never look at that part until an event like this forces them to. Atlacis helps owners map exactly this kind of dependency before it becomes an emergency: which AI tools the business actually relies on, who is really behind each one, and what the fallback looks like if access, ownership, or terms change without warning. The question worth asking about any AI vendor a business is building real workflows around is not just what the tool can do today. It is what happens to the business if that vendor's situation changes, and whether the business would find out in time to do anything about it.

The short version

  • Manus, an AI agent platform Meta agreed to acquire for about $2 billion in December 2025, is returning to independent operation after China's National Development and Reform Commission ordered the deal unwound in April 2026 on national security and foreign-investment grounds.
  • Manus told users on August 11, 2026 that data generated by certain users on or after December 29, 2025 will be deleted to comply with regulatory requirements; affected users have until August 23, 2026 to back up their data.
  • This was not a hack or a security failure. It was a government ordering a corporate ownership change reversed, with real, active consequences for the platform's actual customers.
  • The exposure is not limited to tools with origins in any one country. Any AI vendor whose ownership or funding crosses a border a regulator cares about carries a version of this same risk.
  • Most businesses evaluate an AI vendor on cost and capability and never check who owns the company or whether that ownership is contested, until an event like this forces a scramble with a short deadline.
  • Map which AI vendors your business actually depends on, check who owns each one, and keep independent backups of anything important that lives inside an AI tool, regardless of which vendor you use.
Tags:AI governanceAI vendor riskvendor dependencydata exposureAI buying decisionsbusiness AIAI decision-makingAI regulation
FAQ

Common questions

Was Manus hacked or breached?
No. There is no indication of a security breach. The data deletion is Manus complying with a regulatory order tied to the unwinding of Meta's acquisition, not a security incident.
Does this mean AI tools founded outside the United States are unsafe to use?
No. This dispute is about foreign investment rules and national security review of a specific corporate acquisition, not about the safety or quality of the underlying technology. The broader lesson applies to any AI vendor whose ownership crosses a border a regulator cares about, not to any one country's tools specifically.
What should a business using Manus do right now?
Check whether your account is affected by the August 23, 2026 data deletion deadline and back up anything important immediately if it is. More broadly, treat this as a prompt to check the ownership status of every AI vendor your business depends on and to keep independent backups of important AI-generated work regardless of vendor.
Keep reading

More from the blog

China is discussing limits on overseas access to its own AI models. Here is what business owners using cheap Chinese AI should know.

Reuters reported on July 7, 2026 that Chinese authorities have held meetings with Alibaba, ByteDance, and Z.ai about restricting overseas access to their most advanced AI models, the same low-cost, high-capability models many businesses have adopted to cut rising AI token costs. Nothing has been decided, and any limits may apply only to future models. Here is what business owners should understand about this new kind of vendor risk before it changes anything about how they use AI.

IBM surveyed 1,000 executives on AI vendor risk. 91 percent do not know what they depend on. Here is what that means for your business.

A June 2026 IBM Institute for Business Value study found that 91 percent of senior executives do not fully understand their AI dependencies. The research puts a profit number on that gap. The lesson is not to buy more AI. It is to understand what you already depend on.

The AI vendor default era is over. Here is what business owners should do instead.

For three years, most businesses defaulted to OpenAI. In May 2026, Anthropic overtook OpenAI in US business AI spending for the first time, according to the Ramp AI Index. DeepSeek went from near zero to the fastest-growing measured vendor in the same period. The AI vendor market has fragmented, switching costs are near zero, and the decision is no longer about picking the right tool. It is about building so you can change your mind.

Make better AI decisions, starting with one call.

Book a free AI Fit Call. We will tell you what to use, what to avoid, and where to start. No jargon, no pressure.