In a precedential decision with significant implications for artificial intelligence (AI) and software patents, the Federal Circuit in Ancora Technologies, Inc. v. Roku, Inc., No. 2023-1674 (Fed. Cir. June 16, 2025), reaffirmed that claim terms—such as the term “agent”—will be given their plain and ordinary meaning, encompassing both software and hardware implementations, unless the patent specification or prosecution history contains a clear and explicit disclaimer or definition. The court also clarified that actual license agreements are strong evidence of nonobviousness and need not be tied to specific claims or proven to be the sole motivation for the license.
Key takeaways for patent practitioners
- Broader Claim Scope for “Agent” Terms: The decision confirms that, absent clear limiting language, claim terms like “agent” in AI and software patents will be interpreted broadly to include both software and hardware implementations. Practitioners should be aware that courts will not read in limitations unless expressly stated in the intrinsic record.
- Importance of Clear Definitions: To avoid unintended breadth or narrowness, patent drafters should define key technical terms—especially those central to AI and software inventions—within the specification or make explicit statements during prosecution. This is critical to preserve the intended claim scope in litigation and post-grant proceedings.
- Licensing as Evidence of Nonobviousness: The Federal Circuit emphasized that actual license agreements to a patent are highly probative of nonobviousness. Such agreements do not require parsing to specific claims or proof that the patent was the sole reason for the license. Patent owners should maintain thorough records of licensing activity, as these can serve as powerful objective indicia of validity.
Summary of the decision
The case involved Ancora’s patent directed to methods for restricting unauthorized use of licensed software on computers, specifically through the use of an “agent” to set up a verification structure in a computer’s basic input/output system (BIOS). The Patent Trial and Appeal Board (PTAB) found the claims obvious, construing “agent” broadly as “a software program or routine.” On appeal, the Federal Circuit affirmed this construction, rejecting arguments to limit “agent” to software-only or OS-level implementations, and found no clear disclaimer in the record. The Federal Circuit also upheld PTAB’s obviousness determination but vacated and remanded on the issue of secondary considerations, holding that the PTAB applied an improperly heightened standard to licensing evidence.
Practical implications
- For Patent Prosecution: Practitioners should proactively define or disclaim the scope of key terms in AI and software applications to avoid unintended interpretations. Where broad protection is desired, avoid unnecessary limitations; where narrow protection is intended, make disclaimers explicit.
- For Litigation and Post-Grant Proceedings: Expect courts and the PTAB to apply the plain and ordinary meaning to claim terms unless the record compels otherwise. Licensing history should be leveraged as strong evidence of nonobviousness, even if licenses cover multiple patents or result from litigation settlements.
For more information on the content of this alert, please contact your Nixon Peabody attorney or the authors of this alert.