PatentNext Summary: Sufficiency of disclosure for Artificial Intelligence (AI) inventions in the U.S. can be supported by expert testimony opining on the knowledge that one of ordinary skill in the art would have held based on the disclosure for the patent specification.


In an earlier article, we compared the sufficiency of disclosure for Artificial Intelligence (AI) patents in the U.S. and the European Patent Office (EPO). See A Tale of Two Jurisdictions: Sufficiency of Disclosure for Artificial Intelligence (AI) Patents in the U.S. and the EPO.

In that article, we saw how two applications, one in the U.S. and the other in the EPO, were treated differently despite having the claims. In particular, the U.S. application was prosecuted to issuance, while the EPO application ended in a rejection for lack of sufficiency of disclosure. The article provided a more detailed summary of the EPO case, as a sufficiency of disclosure issue was raised in the EPO case, but not in the U.S.

In this article, we explore how sufficiency of disclosure issues for AI-related claims can arise in the U.S. in the context of a recent district court case. In the U.S., expert opinions and testimony may help a court decide issues of sufficiency of disclosure.

For example, in Centripetal Networks, Inc. v. Cisco Sys., Inc., plaintiff Centripetal Networks, Inc.  accused Cisco Systems, Inc. of infringing several patents “involving complex issues in cybersecurity technology.” 492 F. Supp. 3d 495, 506 (E.D. Va. 2020). One of the patents-at-issue (U.S. Pat. No. 9,500,176)  recited a system “identifying malware-infected computers through use of correlation.” Id. at 549.

Cisco’s accused products included a “Cognitive Threat Analytics (CTA)” software for monitoring a computer network, including monitoring “for security breaches within the network by using machine learning.” Id. at 517.

Cisco argued that the asserted patent was invalid for lack of an adequate written description pursuant to 35 U.S.C. § 112(a). In particular, Cisco, through its expert witness, argued that the asserted patent was invalid “because the [patent’s] specification … contains no description of Cognitive Threat Analytics, machine learning, [or] artificial intelligence …” to demonstrate that the inventors “were in possession of the invention … covered by the scope of the claims … alleged [by plaintiff].” Id. at 558.

The court disagreed. The court looked at the patent specification, in conjunction with Centripetal’s opposing expert testimony, to find that Cisco failed to prove by clear and convincing evidence that the asserted patent lacked sufficient written description. 

In particular, the court found persuasive the testimony of Centripetal’s expert, who opined that a person of ordinary skill in the art would be able to read the patent specification and determine that the invention “utilize[s] logs to correlate packets transmitted by one or more network devices with packets received by one or more network devices” and “identifies hosts associated with malicious entities and communicates messages identifying that host.”  Id. at 558 (citing expert testimony). This “process occurs in response to the correlation of data.” Id. Based on the sections of the specification, the court found that a person skilled in the art would have been in possession of the invention at issue, and, therefore, that the asserted patent was found to not lack sufficient description. Id.

The case serves as an example of how expert testimony, in conjunction with sufficient support in a patent specification, can be used to demonstrate the sufficiency of disclosure in U.S. patent proceedings for AI-related inventions, and, more generally for software-related inventions. For more on the sufficiency of disclosure for software-related inventions, including AI, see: Why including an “Algorithm” is Important for Software Patents (Part 1) and (Part2).

Subscribe to get updates to this post or to receive future posts from PatentNext. Start a discussion or reach out to the author, Ryan Phelan, at or 312-474-6607. Connect with or follow Ryan on LinkedIn.