USPTO Guidance on Use of AI-Based Tools in Practice: How to Mitigate Risk | Wolf, Greenfield & Sacks, P.C.
Use of AI-based tools are permitted, but parties should use caution to ensure compliance with all duties and obligations
Recognizing that AI tools are already being used by parties practicing before the USPTO, this week the USPTO published guidance concerning the use of AI-based tools in practice. The guidance highlights existing rules and obligations, which apply regardless of whether AI tools are used, and indicates instances where use of AI tools may present unique risks. Key takeaways are summarized below. The full text of the guidance can be found here.
Drafting Documents for Submission to the USPTO
Use of AI tools to assist in preparing documents for submission to the USPTO may be advantageous, but parties practicing before the USPTO have an obligation to carefully review the documents prepared using AI tools to ensure the documents comply with USPTO rules and policies.
- A party presenting a paper to the Office certifies under 37 CFR 11.18(b) that “all statements made therein of the party’s own knowledge are true” and “all statements made therein on information and belief are believed to be true.” Parties submitting such documents must ensure the contents of the document complies with these certifications. This is particularly important where generative AI systems are capable of omitting, misstating, or making up (“hallucinating”) information.
- Parties have a duty of inquiry that is “reasonable under the circumstances” to confirm all facts presented in the document “have evidentiary support” or are “likely to have evidentiary support after a reasonable opportunity for further investigation or discovery.” The USPTO makes clear that “[s]imply relying on the accuracy of an AI tool is not a reasonable inquiry.”
- While there is no per se requirement to notify the USPTO regarding the use of AI in preparing submitted documents, disclosure may be nevertheless required in cases when use of the AI tool is material to patentability. Material information could include evidence that no natural person made a significant contribution to a claimed invention because the purported contributions were instead made by an AI tool. This could occur when an AI tool is used to draft the patent application and the AI tool introduces embodiments which are claimed and to which the named human inventor(s) did not contribute significantly. Further explanation of this issue can be found here.
- Parties have a duty to refrain from prosecuting claims which are known to be unpatentable. For example, when an AI tool is used to assist in drafting patent claims, parties have a duty to review and modify the claims to ensure the presented claims are not known to be unpatentable (e.g., due to clear lack of compliance with statutory requirements, such as 35 U.S.C. § 112).
- AI tools may be used to assist in submitting information material to patentability to the USPTO. Use of AI tools to assist in preparing Information Disclosure Statements has the potential risk of increasing the number and size of IDS submissions, which could burden the Office. The party submitting the IDS has a duty to review the information presented in the IDS to ensure the document does not present clearly irrelevant or marginally pertinent cumulative information. Parties must also ensure documents prepared using an AI tool do not inadvertently omit information that is material to patentability.
Utilizing USPTO Resources
Use of AI tools when utilizing USPTO resources must not violate USPTO rules and policies.
- AI tools may be used to assist or automate aspects of filing documents at the USPTO, but parties must ensure that USPTO rules and policies are not violated when using AI tools in this manner. AI tools may not be used to insert a signature of a person on a document submitted to the USPTO. In addition, an AI tool may not obtain a USPTO.gov account and parties may not sponsor an AI tool as a support staff individual to obtain a USPTO.gov account for the AI tool.
- AI tools may be used to interact with USPTO information technology systems, but use of such tools must be monitored to ensure they do not file documents or access information they are not authorized to access. In addition, use of AI tools to mine information from the USPTO may conflict with the Terms of Use for USPTO websites, as use of computer tools which generates unusually high numbers of database accesses is prohibited. The USPTO provides bulk data products that may be used instead.
Confidentiality and National Security Concerns
Parties must carefully vet AI tools before use to ensure an understanding of the AI tool’s terms of use, privacy policies, and cybersecurity practices.
- Care must be taken to ensure that use of AI tools does not result in inadvertent disclosure of client-sensitive or confidential information. For example, when an AI tool is used to perform prior art searches or to assist in drafting documents, information about an invention may be input into the AI tool. Some AI tools may retain such information and it may be shared with third parties intentionally or inadvertently, such as by a data breach.
- Use of AI tools can also implicate national security, export control, and foreign filing license issues. For example, an AI tool may use servers located outside of the US such that data entered into such an AI tool may be exported outside of the US, which may be a violation of certain regulations or orders.
Fraud and Intentional Misconduct
Parties must ensure that actions performed using an AI tool complies with the duty of candor and good faith.
- The guidance also cautions that the duty of candor and good faith, which prohibits fraud and intentional misconduct, extends to the actions individuals perform using AI tools. In addition, use of AI tools on USPTO websites for “[u]nauthorized access, actions, use, modification, or disclosure of the data contained herein or in transit to/from [USPTO web systems] constitutes a violation of the Computer Fraud and Abuse Act.”
Conclusion
This guidance is one of the USPTO’s efforts to address the impact of AI on intellectual property. Guidance on determining inventorship of AI-assisted inventions was previously provided by the USPTO. Additional details on the USPTO’s inventorship guidance can be found here. We expect further guidance on the intersection of AI and IP to be released in the future.
