Beijing Clarifies AI Training System's Legal Position for the First Time: Data Poisoning of Model Classified as Crime of Endangering Computer Information System

According to DynaBeat Monitor, the Beijing People's Procuratorate has released the "Beijing Procuratorial Authority Intellectual Property Prosecution White Paper (2025)," which disclosed that in 2025, the city's procuratorial organs handled 113 intellectual property cases related to the artificial intelligence and data element industry. The white paper highlighted two landmark cases in the AI field.

The Dongcheng District Procuratorate handled Beijing's first case of illegal deletion of AI model training data. An employee intentionally deleted a large amount of data used by the company to train an AI model, causing structural damage to the training system and significant economic losses. The procuratorial authority filed a public prosecution for the crime of destroying computer information systems. The white paper stated that this case helped establish AI model training systems as "computer information systems" in the criminal law sense. This charge previously mainly focused on cases where traditional information systems such as servers and databases were invaded or damaged, and the criminal position of AI training data and training systems was not clear.

The Tongzhou District Procuratorate handled Beijing's first case of copyright infringement using an AI-generated model. Four defendants reproduced others' works using AI technology without authorization for profit, and all were sentenced for copyright infringement.

Throughout 2025, the Beijing procuratorial organs handled a total of 1195 intellectual property cases, a year-on-year increase of 10.34%.