Is AI stock trading software illegal? Do programmers developing such products commit any crimes? A real case analysis from Shanghai

null

With the rapid development of the digital economy, artificial intelligence is undoubtedly one of the most innovative and promising fields today. Many AI entrepreneurs and programmers are actively involved, full of enthusiasm and technical dreams. However, seemingly innovative business models may also conceal numerous unnoticed legal risks.

This case will analyze in depth the legal risks faced by AI entrepreneurs, programmers, and technical teams when engaging in financial technology, quantitative trading, and related fields, through Shanghai’s first case of illegal securities business involving an “AI stock trading robot,” and will offer compliance suggestions.

I. Author of this article: Lawyer Shao Shiwei

1. Shanghai’s first illegal operation case of AI stock trading software: the company’s control person sentenced to seven years and nine months

Case overview:

Company S, without approval from the regulatory authorities, operated the online platform “Xundong Quantitative,” and promoted stock trading products such as “interval arbitrage” and “DIY stock trading robots” to clients, providing securities consulting services like timing of individual stock trades and recommending specific stocks. The total profit amounted to over RMB 30 million[i].

After trial in two levels of courts, the actual controller of Company S, Zhong, was sentenced to seven years and nine months.

2. Is selling AI robot stock trading and quantitative trading software illegal?

In this case, the boss of the tech company, Zhong, believed that[ii]:

His self-developed DIY stock trading robot is based on plans set by clients themselves, helping clients screen and buy stocks they want. In Zhong’s view, the trading strategies are all drafted by clients, and the software merely performs data analysis, with no recommendation of stocks, so the company does not need relevant qualifications for operation.

However, this view was not adopted by the court. Many AI industry practitioners, including programmers, might hold similar opinions:

AI stock trading software is just an information filtering tool, not investment advice. The software automatically captures and organizes public market data (such as capital flows, trading volume, etc.), helping users improve information processing efficiency. The final decision-making power remains entirely with the user, which is fundamentally different from providing specific stock buy/sell recommendations or promising returns (“stock recommendation”). Therefore, they believe such software belongs to “neutral information technology services,” not “illegal securities investment consulting,” and should not constitute illegal operation of securities business as defined in Article 225 of the Criminal Law.

In this view, “factors” (such as capital movements, sector shifts) and “interval arbitrage” models are based on publicly available, objective data for calculation and presentation. The software’s role is merely to simplify information processing, without subjective “value judgments” or “investment decisions.” Users trade based on their own settings, and the software only automatically executes when conditions are triggered, which is an autonomous tool following user decisions, not a company making decisions on behalf of users.

If merely using data analysis and automation technology constitutes a crime, then all similar AI software or financial data terminals (such as Tonghuashun, Eastmoney functions) providing information filtering and quantitative tools could also be deemed illegal operations, which is clearly unreasonable.

So, why was the business of Company S deemed an illegal operation?

3. Why did the court determine it constituted illegal business: analysis of Company S’s business model

From the perspective of Company S’s business model, the actual situation far exceeds simple sales of AI stock trading software. The company not only provided “interval arbitrage” and other strategy services but also categorized its stock trading robot software into different membership tiers, with VIP memberships costing RMB 8,800 and RMB 28,800 respectively. Through these memberships, S charged clients “interface usage fees,” i.e., fees for “illegally accessing brokerage trading channels,” ultimately earning over RMB 3 million.

Specifically, the RMB 8,800 tier members received services like “interval arbitrage.” “Interval arbitrage” refers to software analyzing stock price fluctuations over several days, calculating reference values, and recommending stock buy/sell timing or directly providing trading strategies to assist clients in quantitative investment. If clients are dissatisfied with preset algorithms and strategies, they can seek further analysis and recommendations from customer service.

The RMB 28,800 tier members can set their own investment “tracks” and capital allocation plans based on data, models, and parameters provided by the company. Once preset conditions are triggered, the software automatically helps execute buy/sell operations.

Therefore, the reason for the illegal business charge against Company S is:

The RMB 8,800 service essentially provides specific investment advice—recommending stock buy/sell timing, offering strategic trading, with customer service analyzing and recommending based on experience. This service directly involves “telling clients what to buy and when,” which is a core investment advisory activity, exceeding the scope of information intermediary.

The RMB 28,800 service, by combining company-provided data and models, directly replaces client execution of trades. The software’s automatic execution function essentially acts as a trading channel and instruction execution, which is the core of brokerage business.

This combination allows Company S to bypass licensed brokerage roles, directly completing core securities trading and investment consulting activities that should be performed by licensed institutions like securities firms and investment advisory companies.

4. The compliance boundary of AI stock trading software: distinguishing between criminal and non-criminal conduct

Based on the above case, does it mean that individuals or companies selling AI stock trading software without obtaining qualifications such as the “Securities and Futures Business License” or “Securities Investment Fund Sales Qualification” are automatically violating the law? Of course not.

In practice, the core controversy is not whether AI or quantitative methods are used, but whether the final product and service output are data and tools, or “investment advice and trade execution.” In other words, whether the service has crossed from neutral technical support into the boundary of securities investment consulting or securities business operation.

Considering Company S’s business model, the comparison can be made at three levels:

1. Function output level:

In S’s services, the “interval arbitrage” model and customer recommendations produce clear instructions on “when to buy or sell which stock.” This completes the transition from “what” (data facts) to “how” (investment decision), which is the essence of investment consulting.

According to the “Interim Regulations on Strengthening Supervision of Securities Investment Consulting Business Using ‘Stock Recommendation Software’ (2020 Second Revision),” software products, tools, or terminal devices that have functions such as aggregating securities information or statistical analysis of securities investment varieties’ historical data, but do not have functions like “providing specific securities investment analysis opinions, predicting price trends, offering variety selection suggestions, or actual buy/sell recommendations,” are not classified as “stock recommendation software.”

Therefore, if an AI tool only provides securities information aggregation, historical data statistics, capital flow or sector shifts, and does not offer specific analysis opinions, trend predictions, variety suggestions, or buy/sell advice, it is closer to a neutral information tool.

2. Business logic level:

Company S’s business model involves charging high, tiered membership fees based on “providing profitable investment advice and trading convenience.” From the user’s perspective, they are paying for “who can help me quickly make money.” Essentially, this is selling investment consulting and trading channel services.

In contrast, if the product is only a one-time sale of software or charges a data subscription fee, with the main value being the information tool itself, and not emphasizing “profitability” as the core selling point, it is more aligned with a technical or information service business logic.

3. Operational loop level:

In S’s services, especially the RMB 28,800 tier, users set conditions, and the software fully automates the process from decision trigger to order execution. This partially replaces the roles of investors and securities brokers.

In contrast, if the product only provides analysis tools, and users need to judge and execute trades themselves, with decision and execution not automated by the platform, the overall risk boundary is more controllable.

5. “Black technology” enhancement leading to multiple penalties

The reason S’s software can quickly assist clients in stock automation trading is that it purchased an external hacking program from “hacker” Han, integrated it into its own system, and deployed it on company servers. This allowed S’s account to bypass the technical protections of Tongdaxin (a market data terminal provider), access trading channels illegally, and perform automated stock trading, charging clients interface usage fees.

Han was sentenced to three years for providing intrusion into computer information systems. Zhong, the actual controller of S, and the company’s technical director, Kong, were sentenced for multiple crimes including illegal operation and copyright infringement.

The reason is that, within a compliant framework, the basic structure of securities trading should be:

(The person still prefers the original diagram style)

Exchanges are responsible for trading rules and matching mechanisms; securities firms, as the sole legitimate operators, establish client relationships with investors; investors participate in securities trading through securities firms. Market data and trading terminal service providers like Tongdaxin support technical tools for securities firms and investors, but do not engage in securities business themselves nor establish securities service relationships with investors.

However, in this case, due to S’s involvement, the structure evolved into:

Investors first establish trading relationships with S, purchasing products and services like “interval arbitrage” and “stock trading robots.” S then cracked the Tongdaxin interface program, bypassed normal protections, illegally accessed brokerage trading channels, and completed automated trading, continuing to charge investors fees.

Under this structure, securities firms and Tongdaxin are not organizers or beneficiaries of the business but are passively exploited as “targets” of the technical pathway. The true entities that build the business model, control the trading path, and generate revenue are always S itself.

6. Risk warning for AI entrepreneurs and programmers:

This case of a tech company being sentenced for “AI stock trading software” essentially exposes the ambiguity in the compliance boundaries of many AI startup projects, quantitative trading products, and fintech tools.

For AI entrepreneurs, company leaders, technical partners, programmers, and product managers, if you are engaged in or planning to enter related fields, it is recommended to carefully evaluate your products and business models to ensure they do not cross the regulatory boundaries of securities business, avoiding misclassification as investment consulting or disguised securities operation.