April 10 (Reuters) - The U.S. Patent and Trademark Office on Wednesday told lawyers seeking patents that they must disclose any "material" part played by artificial intelligence in creating an invention.
The office issued new guidelines cautioning patent attorneys against omitting AI involvement from their patent applications and using AI to draft documents without verifying the underlying information, which it said could result in "critical misstatements and omissions."
USPTO director Kathi Vidal said in a statement that the guidance was "part of our work shaping AI policy" and "encourages the safe and responsible use of AI to benefit the IP and innovation ecosystem."
The new guidance follows the Biden administration's order last year for federal agencies to address emerging artificial-intelligence technology.
The USPTO said in separate guidance in February that it will only grant patents for AI-assisted inventions that feature "significant" human contributions.
The office elaborated on Wednesday, in a document set to be published on the Federal Register on Thursday, that if there is a question about the extent of human contribution to an invention, "information regarding the interaction with the AI system (e.g., the inputs/outputs of the AI system) could be material and, if so, should be submitted to the USPTO."
The office also said that patent and trademark lawyers must confirm the accuracy of documents created with AI assistance before submitting them, "given the potential for generative AI systems to omit, misstate, or even 'hallucinate' or 'confabulate' information."
It noted that attorneys were previously sanctioned in New York for submitting a legal brief with fake AI-generated case citations.
The USPTO also warned patent attorneys to be careful when entering client data into AI systems for assistance on patent applications, which the office said could raise confidentiality concerns, as well as national security issues if the systems are located in foreign countries.
Comment