In this case disclosed by the Xuhui District People's Procuratorate in Shanghai, the "sheep-shearing script" has for the first time targeted the large model computing power itself. Zong XX first used AI to generate and assist in writing automated scripts, registering a large number of new accounts on a certain AI platform, falsifying electronic signatures and circumventing IP restrictions, thereby stealing a bulk of the rewards points originally intended for attracting new users; subsequently, he discovered that a certain large model aggregation platform had undisclosed interfaces, which he connected to a self-built intermediary website to provide low-priced AI drawing and video generation services, while the actual cost was borne by the original platform's paid computing power that was "stolen" from behind the scenes. It has been reported that the relevant interfaces were called approximately 290,000 times, and Zong XX illegally profited around 34,000 RMB. Ultimately, the Xuhui District Procuratorate prosecuted him for illegally obtaining data from a computer information system, and the Xuhui District Court sentenced him to three years of imprisonment, three years of probation, and a fine of 40,000 RMB; around November 3, 2025, details of the case were disclosed to the public, making it the first criminal case in Shanghai involving the use of AI to falsify signatures and misappropriate large model computing power. Unlike traditional paths of theft and fraud, this case incorporated "computing power + interface calling rights" into the category of computer information system data protection, drawing a new criminal red line for AI platforms and Web3-like aggregation services reliant on interface aggregation, incentive points, and computing power transit: script brushing, point farming, and privately accessing internal network interfaces are no longer merely "technical grey areas" but are now direct issues of compliance with criminal liability and platform risk control boundaries.
AI Script Point Farming: How Computing Power Was Moved to the Black Market
Before detailing specific operations, Zong XX first industrialized "script sheep-shearing" using AI. The prosecution disclosed that he did not manually register accounts but instead used AI to generate or assist in writing automated scripts that allowed the program to batch fill registration information, falsify electronic signatures, and circumvent the electronic signature verification and IP registration restrictions that were originally meant to validate authenticity on the platform. Every "new user" could obtain points or calling quotas distributed by the platform, and the script continuously produced accounts like an assembly line in a factory, turning the point pool meant for attracting new users into his own "supply channel," while the platform’s points and calling quotas were treated as digital assets that could be extracted at no cost during this process.
What truly dragged computing power into the black market was his discovery of an undisclosed interface on a certain large model aggregation platform. This interface, which should have been hidden within the platform's internal calling system, was integrated into his self-built intermediary website, packaging it as a low-cost AI drawing and video generation service. The front end charged users a small fee while in the background, it called upon the original platform’s paid computing power resources at an alarming rate. According to reports, this relevant interface was illegally called approximately 290,000 times, bringing him about 34,000 RMB in illegal profits, with nearly all computing power costs "outsourced" to the aggregation platform, while he was only responsible for peddling the cheap services and pocketing the intermediate price difference. In this chain, the calling rights and point quotas were broken down into three phases: procurement, processing, and retail, thus transforming computing power from an operational cost on the platform’s balance sheet into a "black market product" that could be arbitraged and resold.
Abuse of Undisclosed Interfaces: Platform Security Boundaries Torn Apart
From the case details, it is evident that Zong XX did not target the official entrances listed in public documents but rather the "undisclosed interfaces" that existed within the internal structures of the large model aggregation platform. In theory, such interfaces should either be open only to specific business entities or secured through strong authentication, whitelists, and granular permissions, yet the reality is that they were integrated into a self-built intermediary website over a prolonged period and reportedly called around 290,000 times, while the platform failed to identify and block them. This indicates a near loss of control over the lifecycle management of the interfaces: the interfaces were not included in unified governance, access control was practically non-existent, and calling logs as well as abnormal traffic alerts did not serve their purpose as a safety net. The result was that an entrance the platform believed was "not open to the public" turned into a weak point that was prioritized for discovery by scripts, crawlers, and grey industries, thus tearing apart the security boundary from its most vulnerable layer.
A more profound contradiction lies between the platform’s commercial incentives and technical protections. While the platform distributes points or free calling quotas to new users to attract them, it lacks accompanying mechanisms to prevent scripts from being used for bulk registration and sheep-shearing, effectively pushing computing power subsidies towards "the most diligent programs." In cryptocurrency exchanges, quantitative platforms, and DeFi scenarios, API calls have already been viewed as high-risk entrances, typically requiring additional API keys, IP whitelists, rate limits, and behavioral risk controls. Operators would often include these control measures in user agreements and compliance disclosures to demonstrate that they had fulfilled necessary security obligations. However, this case exposes the AI computing power platform's significant lag in managing interfaces that are also "high-frequency automated calls." The Cybersecurity Law and Data Security Law require network operators to adopt technical and management measures to ensure system and data security, meaning that interface design, incentive mechanisms, and risk control systems are also part of compliance boundaries: when platforms use free quotas to exchange for growth while allowing undisclosed interfaces to run free for extended periods, the risk is no longer just a technical vulnerability but could directly become a target for regulatory and judicial scrutiny in similar future cases.
From Account Theft to Computing Power Theft: Criminalizing Illegal Access to System Data
In this case, Zong XX's actions superficially appeared to be "farming points and reselling computing power," but the court did not follow the traditional paths of theft or fraud. Instead, it clearly convicted him of illegally obtaining data from a computer information system, and this choice itself serves as a signal. If handled as theft or a contract dispute, the focal point would fall on "whether platform points constitute property" and "whether the rules for attracting new users are unclear," easily being interpreted as a commercial dispute centered around promotional terms. However, in this case, Zong XX utilized AI to write scripts for the bulk registration of accounts, falsified electronic signatures, circumvented IP restrictions, and then discovered and accessed the platform’s undisclosed interface, providing low-priced services through the intermediary website while substantively calling upon paid computing power. The core issue has shifted from "Did he take the money?" to "Did he bypass authentication and scale up the acquisition of data and resources under the system's control?" The court prioritized the protection of the platform's system data and operational order and chose to respond to this automated, large-scale "sheep-shearing" using computer crime provisions.
To understand this classification, the key lies in what is considered "computer information system data." The understanding of this concept in Chinese criminal law and relevant judicial interpretations is already quite broad, encompassing not only accounts, passwords, and authentication data but also system-generated and recorded points, interface calling logs, billing records, etc. In Zong XX’s operational chain, the accounts and login credentials obtained via bulk registration, the new user points distributed and automatically computed by the system, the logs and billing data generated from approximately 290,000 unauthorized interface calls, as well as the corresponding usage rights of computing power all effectively belong to the same set of computer information systems and constitute a "data-profit" whole that can be illegally obtained and utilized. The court does not dissect this into "Are points money?" and "Is computing power a commodity?" but rather views it as an illegal acquisition and control of system data and its extended benefits in total, logically falling under the framework of illegally obtaining data from a computer information system. It is no wonder that the media interpreted it as a typical case categorizing computing power resources and relevant data rights under criminal law protection.
This path will have a chain effect on the future criminal regulatory boundary for computing power, API quotas, and even on-chain resources. Once judicial practice regards "account systems + quotas/allocations + calling records + usage rights" as an integral data asset and digital usage right, actions that bypass authorization to bulk extract and resell these resources will become increasingly difficult to package as mere contractual disputes or "grey arbitrage." Whether it be the paid calling quotas of AI platforms or API keys and access quotas possessed by aggregation service providers, or even digital resources allocated through system parameters like bandwidth and Gas limits on public chains, when they are structurally abused and resold through "shadow gateways," they could all potentially fall under the evaluation framework of "illegally obtaining system data." For all participants relying on computing power pools, API quotas, or on-chain resources to conduct business, this case marks a new coordinate in criminal regulation: bypassing authorized extraction and reselling "digital usage rights" has been viewed by the judicial system as a high-risk behavior liable to criminal penalties.
A Triple Warning to AI Platforms, Exchanges, and Developers
For AI platforms and large model aggregators, this case first breaks through two "illusions": firstly, Zong XX's large-scale calls were to the platform's undisclosed interfaces, yet the court still determined it constituted illegal acquisition of computer information system data, indicating that "not documented or publicly promoted" does not equate to effective access control; secondly, while the platform provides points or calling quotas to new users for attraction, it lacks mechanisms to prevent scripts from being used for bulk registration and sheep-shearing, ultimately being bypassed for electronic signature verifications and IP restrictions, transforming the event into one evaluated under criminal law. Combined with the requirements of the Cybersecurity Law and the Data Security Law regarding the safety obligations of network operators, this means that AI platforms and aggregators must treat interface management and regulatory compliance as a "core business" rather than a supplementary measure: unauthorized calls must have technical barriers, logs must be traceable, and there must be clearly defined rules; otherwise, alongside user behaviors leading to criminal charges, the platform may also be held accountable at civil and administrative levels.
If the perspective is expanded to cryptocurrency exchanges, computing power markets, and Web3 developers, this case provides another layer of warning: the seemingly "farming points" model is not fundamentally different from exchanges transporting traffic through APIs, harvesting fees and commissions, or abusing recharge rewards; decentralized computing power markets and AI+Web3 projects that utilize smart contracts layered with APIs to provide computing power and model calls, once bypassing risk control through scripting, falsifying identities, or exceeding authorization to call, and then repackaging quotas for low-priced resale, could be compared to the illegal acquisition of system data, disrupting the operational order of the system, and would no longer be considered merely "grey arbitrage." For developers, utilizing AI to generate scripts is not inherently illegal; the red line rests on whether one intentionally circumvents platform verification mechanisms, knowingly calls in bulk without authorization, and whether one resells usage rights that are originally limited to their accounts for profit; within this coordinate system, automated tools shift from efficiency enhancers to risk amplifiers, and the truly feasible route is only one: during script design and business model development, treat platform usage terms, interface authorization scope, and the regulatory protection boundaries for "system data" as hard constraints rather than as leverage for post-factum games.
Computing Power Becomes "Assets": Where Will the Next Round of Regulation Focus?
From the first case in Xuhui, the court, through the conviction of illegally obtaining data from a computer information system, focuses on "who controls system resources and who uses them improperly." The quotas of computing power, interface calling rights, and backend operational data are collectively included in the resource pool requiring criminal regulatory protection, which serves as a clear signal for all platforms relying on computing power for attraction. Moving forward, AI platforms and aggregators will have to manage free points, test quotas, and even undisclosed internal interfaces as "assets" with both price and risk: for identity verification, moving from simple phone number registration to more refined device fingerprints, IP profiling, and behavioral feature cross-verification; in risk control modeling, including abnormal bulk registration, abnormal calling frequencies, and "digging holes" behavior within real-time monitoring; at the compliance level, they will need to draft policies on the logging of computing power calls, authorization boundaries, and abnormal reporting mechanisms that can withstand regulatory scrutiny under the Cybersecurity Law, the Data Security Law, and other frameworks. As AI intertwines increasingly with cryptocurrency businesses and computing power trading markets, domestic and foreign regulatory bodies have begun to regard the concentration of computing power, data misuse, and interface security as part of systemic risks. Under cross-chain and cross-border calls, the issue of who is responsible for maintaining logs and who is responsible for damage control will compel platforms to increase investments in KYC, log retention, and compliance node integration. However, the boundaries between computing power and data rights have not yet been clearly delineated: what scenarios still constitute simple breaches of contract, at what points will this escalate to criminal risks of "illegally obtaining system data," how to define usage rights for resale within contracts, and how to inform developers and ordinary users of these invisible red lines all contain significant uncertainties. Further clarification on when computing power will be explicitly included in legal provisions and how the boundaries of interface calling rights will be delineated more intricately will determine how far platforms and developers can advance in the next round.
Join our community for discussions and to become stronger together!
Exclusive Hyperliquid benefits for AiCoin: https://app.hyperliquid.xyz/join/AICOIN88
Exclusive Aster benefits for AiCoin: https://www.asterdex.com/zh-CN/referral/9C50e2
On-chain Telegram community: https://t.me/AiCoinWhaleData
On-chain community: https://www.aicoin.com/link/chat?cid=N6OVMor5g
On-chain Twitter for AiCoin: https://x.com/aicoinwhaledata
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
