China warns of risks in using OpenClaw AI agent for patent applications
China's State Intellectual Property Office warned that OpenClaw and similar AI agent tools pose multiple risks when used to draft patent applications, including technology information leakage, hallucination-induced defects, and dishonest patent practices that could result in administrative penalties for applicants and agencies.
What happened
On April 1, 2026, China's State Intellectual Property Office issued a risk warning about using OpenClaw, an AI agent tool, to draft patent application documents. The warning stated that OpenClaw and similar intelligent agent tools have weak default security configurations and can pose serious safety risks.
The office identified three specific risks: technology information leakage due to excessive permissions, security vulnerabilities and plugin poisoning; substantial defects from AI hallucination causing logical contradictions and unclear technical descriptions; and dishonest patent applications when the AI fabricates, randomly invents or pieces together content.
Reported impact
If technical information is leaked through OpenClaw, patent applications may lose novelty and fail to receive protection, or technology could be patented first by others, causing major losses to applicants. Patent agencies using such tools would also face breach of contract liability.
Applicants engaging in dishonest patent applications face warnings and fines up to 100,000 yuan. Patent agencies and agents risk revocation of their practice licenses and cancellation of their agency qualifications, with serious violations resulting in inclusion in the severe breach of trust blacklist.
Index analysis
The State Intellectual Property Office advised applicants to increase risk awareness, carefully select compliant patent agency services, and actively confirm whether their agent uses intelligent agents to draft application documents. Applicants may lodge complaints and request compensation if they discover unauthorized use by agencies causing information leakage or constituting dishonest applications.
Governance controls
Patent agencies and agents were instructed to be highly vigilant about intelligent agent use risks and eliminate dishonest patent application practices to protect their clients' legitimate interests.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.