On March 31, 2026, China’s National Intellectual Property Administration (CNIPA) issued a warning against using AI agents, such an OpenClaw, in drafting patent application documents (关于使用OpenClaw等智能体撰写专利申请文件的风险提示). Further, unauthorized use by patent agencies should be reported. CNIPA says using agents may lead to technical information leakage, substantial defects including AI hallucinations, and dishonest (bad faith) applications.
A translation of the warning follows. The original text is available here (Chinese only).
Recently, intelligent agent tools such as OpenClaw (formerly known as Clawdbot and Moltbot) have been exposed to have weak default security configurations, which can easily lead to serious security risks. At the same time, using such intelligent agents to write patent application documents may also induce multiple risks. The following are some points to note:
1. Risk of “Technical Information Leakage”: Intelligent agents such as OpenClaw have risks such as excessive privileges, security vulnerabilities, and the potential for malicious plugins. Using them to draft application documents can easily lead to the leakage of core information such as technical disclosure documents. Once leaked, the technical solution of the patent application may be unable to be authorized due to loss of novelty, or even be preemptively patented by others, causing significant losses to the applicant. The agency will also be liable for breach of contract .
2. Risk of “substantial defects”: When using such intelligent agents to draft application documents, “ AI hallucinations” may occur, resulting in problems such as logical inconsistencies in the content of the application documents and unclear descriptions of technical features, thus failing to obtain protection.
3. Risks of “Dishonest Applications”: Creating patent applications through methods such as generating artificial intelligence agents out of thin air, random fabrication, or content patchwork constitutes dishonest patent application behavior that violates the principle of good faith. If a certain number of such applications are submitted, the applicant will face administrative penalties such as warnings and fines; the agency and agent will face administrative penalties such as revocation of their practice license and cancellation of their agency qualification certificate, and in serious cases, they will be included in the list of seriously dishonest and law-breaking entities.
To prevent the aforementioned risks and safeguard the legitimate rights and interests of all parties, the following reminders are provided:
1. Applicants: Applicants should enhance their risk awareness, carefully select compliant patent agency services, and proactively understand and confirm whether the agency uses intelligent agents to draft application documents. If it is discovered that the agency has used related tools without authorization, resulting in information leakage or constituting dishonest application behavior, applicants can file a complaint and demand compensation from the agency for losses in accordance with the law.
2. Patent agencies and agents: They must be highly vigilant about the risks of using intelligent agents , refrain from using intelligent agents to conduct dishonest patent applications, and effectively protect the legitimate rights and interests of their clients.
