China warns of risks in using AI tools for patent applications

China’s National Intellectual Property Administration (CNIPA) has issued a critical risk alert regarding the use of AI-powered tools, such as OpenClaw, for patent application processes. The agency warns that these tools may have default security vulnerabilities that could expose sensitive information, thereby leading to significant risks for patent applicants.

In a statement posted on its official WeChat account, the CNIPA elaborated on the risks associated with agentic AI technologies. Specifically, it highlighted issues including excessive permissions and potential for plugin poisoning, which could compromise the integrity of patent applications.

The CNIPA warns that using AI tools for drafting patent application documents can result in the leakage of essential technical details. Such leaks could jeopardize the novelty of the patent, potentially making it ineligible for approval, and expose applicants to serious losses if others file competing claims.

Additionally, the use of these AI tools may introduce problems known as AI hallucinations. Such occurrences can lead to logical inconsistencies or vague descriptions within applications, further diminishing the chances of securing protection from intellectual property authorities.

According to the CNIPA, those who utilize intelligent agents to create or compile patent application content could potentially engage in bad-faith practices. Serious infractions could lead to administrative penalties, including fines or warnings for both applicants and patent agencies, with grave consequences for repeated violations.

In harsher cases, offenders may find themselves listed as untrustworthy entities, a measure that could significantly impact their professional credibility. The CNIPA emphasizes the need for applicants to be vigilant and informed regarding their choice of patent agency services.

To mitigate these risks, the CNIPA advises patent applicants to carefully choose compliant agency services and to verify their practices. Applicants have the right to take legal action if they discover unauthorized use of AI tools that leads to information leaks or other unethical behaviors.

Furthermore, patent agencies and agents are urged to maintain strict vigilance against the risks associated with AI tools. They must avoid engaging in practices that could undermine the legitimate rights and interests of their clients, ensuring ethical compliance in their operations.

While the adoption of AI agents such as OpenClaw continues to grow, the CNIPA, alongside other Chinese government bodies, has expressed concerns over potential hidden risks that could arise from these technologies.

On March 13, the National Cybersecurity Notification Center issued a caution concerning OpenClaw, alerting users to numerous exposed assets that pose considerable security risks and may become targets for cyberattacks.