Think Before You Chat: Could Disclosing Your Invention to AI Constitute a Public Disclosure?
The United States Patent and Trademark Office (USPTO) is no stranger to the rapidly expanding role of artificial intelligence (AI) in legal practice. From AI-assisted prior art searches to automated drafting tools, artificial intelligence has become increasingly embedded in modern patent prosecution. Yet as inventors and practitioners increasingly turn to generative AI for assistance, a critical question emerges: could disclosing your invention to an AI system constitute a public disclosure that jeopardizes your potential patent rights? Unfortunately, the answer to this critical question is nuanced. While the USPTO has provided some guidance, recent case law and statutory obligations seem to suggest that inventors and practitioners must proceed with caution when utilizing generative AI.
On April 11, 2024, the USPTO issued its “Guidance on Use of Artificial Intelligence-Based Tools in Practice Before the United States Patent and Trademark Office.” In its guidance, the Office acknowledged both the utility and risks of AI tools. But the USPTO also emphasized that it is the individual parties (practitioner and inventor) who remain responsible for reviewing and verifying AI-generated content. Parties submitting AI-generated content must be aware of and cognizant that AI-generated inaccuracies may result in material misstatements or omissions before the USPTO that can result in abandonment of the application or may have a negative effect on the potential patent rights that might be granted. The USPTO noted however that there is no general duty to disclose AI use in drafting unless it is specifically requested. However, complications arise for practitioners and inventors when the use of generative AI interferes with the duty to disclose all information material to patentability to the USPTO.
Any individual associated with the filing and prosecution of a patent application owes the USPTO a duty of candor and good faith. Under 37 C.F.R. § 1.56(a), this duty includes a requirement to disclose all information known to be material to patentability. A broad reading of this duty suggests that if the use of AI is material to patentability, then its use must be disclosed to the patent office. But how does one know if the information is material to patentability? The USPTO provided a telling example to illustrate to practitioners, inventors, and the like, where the duty to disclose information might draw its line.
According to the USPTO, if an AI system drafts patent claims and an individual covered under 37 C.F.R. 1.56(c) knows that a claim lacked significant human contribution, that fact must be disclosed to the patent office. In other words, inventorship issues arising from AI generated contributions may trigger disclosure obligations. The USPTO further clarified that while AI tools are being used internally – for example, examiners employ AI-enable prior art search tools within the Patent End-to-End(PE2E) system – external users must still ensure compliance with certification requirements. Overall, the USPTO guiding principle is clear: AI assistance does not negate a human’s responsibility to disclose information material to patentability to the patent office.
So how does one safely use AI while satisfying their duty of disclosure to the USPTO? First, know the terms and conditions of the AI platform that you are using. If an AI’s terms permit unrestricted data use without confidentiality protections, disclosure of the invention to that system may risk being characterized as public, or in turn may become public if the data is disseminated without restriction. Overall, make sure to use a confidential or private AI system that does not disseminate the information it learns from the user. Second, avoid uploading core inventive concepts prior to filing an application on your idea with the patent office. Once an invention becomes public knowledge, an individual only has one year to file with the USPTO and waive the public disclosure under 35 U.S.C. 102(b). Third, independently verify all AI-generated content for accuracy and validity. Guaranteeing the accuracy of AI generated information will increase the likelihood of compliance with the duty of candor and good faith. Lastly, assess whether the AI’s involvement has affected inventorship in any way. If it looks like the AI has contributed to any claims or embodiments of the invention where the inventor did not originally contribute, then the duty of disclosure might have been triggered and it is better to disclose this information to the patent office than to violate your duty of candor.
Artificial intelligence is not inherently incompatible with patent practice, in fact AI is rapidly becoming a standard tool, but it must be used wisely. The convenience of AI must not obscure the underlying patentability requirements already in place. Current public disclosure doctrine operates on accessibility to information, not the intent on which said information was accessed. Until clearer judicial boundaries emerge, inventors, practitioners, and the like, would be wise to pause before inputting unpublished invention details into any generative AI system. The question is no longer whether AI can help draft your patent application, as it most definitely can. Instead, the real question practitioners and inventors should be asking themselves about using generative AI is: did you accidently disclose more than you intended? Thus, it is always wise to think before you chat.
If you have questions regarding AI and public disclosure of your invention, we encourage you to contact one of our attorneys.
