Detained for 37 days, the first group of people who made their fortune through the "AI Transit Station" have started to get involved.

null

Author: Lawyer Shao Shiwei

According to insiders in the AI industry, in May 2026, the head of a certain AI transfer station publicly issued a statement, claiming that he was detained for 37 days for illegal reverse crawling and reselling low-cost AI interface resources, and is currently under bail pending trial.

Although Lawyer Shao also saw some colleagues in the legal profession stating that the head of an AI transfer station was detained by Shanghai police, since there has been no official announcement so far, the specific case details and handling situation cannot be confirmed.

In the past two years, domestic AI application demand has exploded, but because overseas large models have regional restrictions, more and more people are doing AI transfer stations. Simply put, domestic users want to use but cannot, and transfer stations help open channels, charging a toll.

Therefore, through this case, we can discuss whether the booming AI transfer station business can still be done? What risks might ordinary people face when operating an AI transfer station?

Profitable Business of AI Transfer Stations

According to Xinhua News Agency, from early 2024 to March 2026, China's daily token call volume has increased over a thousand times. The demand is evident.

But if domestic users want to access overseas large models (OpenAI, Anthropic, Google, etc.), they will encounter many hurdles such as network environment, payment channels, and identity verification.

Where there are barriers, there is business opportunity. AI transfer stations have emerged accordingly. Now, on many social media platforms, it is claimed that AI API transfer stations are among the most profitable projects in 2026, which is not untrue.

AI transfer stations can also be understood as AI scalpers. They package interfaces from different AI model providers into a unified outlet, handling all model connections in the background for users. Users also save trouble with VPNs and foreign currency payments.

On Taobao, Xianyu, Xiaohongshu, many such posts can be seen, with prices ridiculously low.

So here’s the question: with prices so low, how do transfer station operators actually make money?

Reselling Free Quota. Platforms like ChatGPT and Claude give free credits when registering new accounts. The behind-the-scenes account merchants register large quantities of accounts, exploit the platform’s free quotas, and use technical means to reverse engineer these accounts’ web interfaces into standard APIs, then sell them externally, with almost zero cost.

Refund Arbitrage. Batch register official accounts, recharge, call APIs. What if accounts get banned? Request refunds. In most cases, pre-paid money can be recovered. Essentially, they use your money to call the API, get banned, then recover costs from the official side—double profit.

Token Overstatement. Official APIs charge strictly based on token count, but the transfer station’s billing system is self-made. Normally, one Chinese character is about 1.5 to 2 tokens, but some transfer stations increase the multiplier in the backend, so one character might be charged as if it used 3 to 4 tokens. Users cannot verify this.

Model Swapping. You buy Claude Opus 4.7, but what’s actually called might be an open-source small model. That’s why many users feel that models used via transfer stations seem “dumber.”

Data Reselling. The transfer station packages and sells users’ complete conversation records, especially high-quality training data like code snippets, reasoning processes, and engineering decisions in programming scenarios, to model vendors. Why are transfer stations cheap? Because they profit from selling data.

This business has grown so large that even celebrities are entering the field. On May 1, 2026, TRON founder Sun Yuchen launched B.AI, an AI transfer station with the slogan “One API Key = Claude + GPT + Gemini + full series of domestic large models.” On May 5, WLFI, a cryptocurrency company linked to the Trump family, launched WorldRouter, directly integrating AI calls with the cryptocurrency system.

But the more traffic, the greater the risks.

Why Are AI Transfer Stations Being Detained?

Previously, we outlined the profit models of AI transfer stations. As operators, they may also be aware of the risks involved in such gray-area projects, but after making money over time and seeing peers doing the same without issues, they gradually relax their vigilance.

From a legal perspective, the criminal risks of AI transfer stations mainly focus on three aspects.

First, the business model itself may be illegal.

AI transfer stations’ computing resources are not obtained through legitimate channels for API interfaces but by bulk registering accounts to exploit free quotas or reverse engineering interfaces via technical means. This no longer falls within normal commercial agency.

Providing information transfer and data processing services is essentially a value-added telecom service. According to the “Telecommunications Regulations of the People’s Republic of China,” operating such services requires obtaining relevant administrative permits. Operating without permission risks violating illegal business operations.

Additionally, overseas large model providers impose access restrictions on users in China. Transfer stations help users bypass these restrictions via proxy IPs, fictitious identities, etc., effectively assisting in circumventing the service provider’s access conditions. If deemed to disrupt market order, such behavior could also fall under illegal business operations.

Second, the lack of data security obligations.

AI transfer stations handle large amounts of user-model interaction data daily. User prompts, code snippets, business documents—all pass through the transfer station’s servers. As the actual data handler, the station bears corresponding security management responsibilities.

However, most transfer stations have not established any data security management system—storage locations, access controls, security measures are all blank. If a data leak occurs, whether due to external attacks or internal mismanagement, the transfer station as a network service provider could face criminal charges for failing to fulfill information network security obligations. This crime targets “failing to perform legally required duties.”

Third, illegal collection and sale of user data.

Some transfer stations package and sell user conversation records to third parties, which is not uncommon in the industry. But the legal risks associated with this are often underestimated.

User interactions with AI models often contain personal information, trade secrets, and other sensitive data. Transfer stations rarely obtain explicit user consent when collecting this data, nor do they fulfill obligations to inform users of data use and flow. Collecting and providing such information to third parties without consent can constitute infringement of citizens’ personal information rights.

The threshold for this crime is not high. According to judicial interpretations, illegally obtaining, selling, or providing more than fifty pieces of tracking, communication, credit, or property information, or over five hundred pieces of personal information such as accommodation, communication records, health data, or transaction info that could impact personal safety or property, can meet the criteria for prosecution. Given the daily data volume processed by transfer stations, reaching this threshold is not difficult.

In Conclusion

Regarding the issue of AI transfer stations, Lawyer Shao does not want to limit the discussion to whether station operators will be prosecuted. Instead, this incident reflects the inevitable problems faced by the AI industry during its rapid development.

For users, transfer stations lower the usage barriers but also expose sensitive data to an unlicensed, unsecured middleman. If problems occur, it may be impossible to find the rights holder to claim remedies.

For vendors, the existence of transfer stations consumes their technological investment and business models. Free quotas are exploited in bulk, paid interfaces are reverse-engineered, and pricing systems are bypassed. Vendors have to divert significant resources from product R&D to risk control, costs that are ultimately passed on to legitimate paying users. More deeply, when transfer stations dump computing power at low prices, the market’s perception of AI service value is distorted—users gradually believe these capabilities should be nearly free. This damages the industry’s sustainable development.

As a lawyer focused on the new economy, Lawyer Shao has been paying close attention to the development of the AI industry and has served many practitioners in this field. How far an industry can go depends not on how fast it runs, but on whether it can establish basic commercial order and trust. The AI industry is at a critical stage from wild growth to regulated operation, and every practitioner’s choices are shaping its future ecosystem.

A healthy AI industry requires continuous technological R&D investment from vendors, genuine protection of users’ data rights, and responsible, compliant participation from practitioners. These are prerequisites for the industry’s long-term survival. Lawyer Shao hopes to see more industry players choosing to do difficult but right things, making the industry’s foundation more solid.

Special Statement: This article is an original work by Lawyer Shao Shiwei and reflects only the author’s personal views. It does not constitute legal advice or legal opinions on any specific matter.