Chantal Whitfield

The Federal Circuit Panel's April 16, 2026 decision in Teva v. Eli Lilly reinstates a $176.5M verdict and draws a clear distinction under §112 between method-of-treatment and composition claims. The decision highlights the importance of drafting choices, like those made by Teva 2006.

Teva shows how much §112 outcomes turn on drafting choices rather than the volume of disclosure. From a specification describing a single humanized antibody, the patentee framed the invention as a method of treatment, anchored the genus to a known biological target, and relied on humanization as routine art within the POSA's knowledge. These are structural choices, made years before litigation, that shape how a patent reads in enforcement decades later.

These choices are exactly what purpose-built tools like Solve Intelligence are designed to surface early on, during drafting, when they can be considered consciously and proactively. The alternative may be that issues only become apparent during litigation, when it is often too late to change a decision that was made previously, whether intentionally or not.

Key takeaways

• Teva v. Eli Lilly (Fed. Cir. Apr. 16, 2026) marks the first clear limit on Amgen v. Sanofi.
• Amgen's §112 calculus does not extend to method-of-treatment claims reciting an antibody genus.
• Same specification, same genus, yet the antibody claims failed in IPR while the headache claims won at trial.
• One humanized antibody, a known biological target, and routine humanization techniques were enough to satisfy §112.
• Four drafting choices won the case: claim category, functional anchoring, routine art, one embodiment.
• Teva is a re-opening event for life sciences portfolios with genus claims.

The leading in-house and outside counsel life sciences patent teams have voted Solve Intelligence as the winner of the Technology Solution of the Year Award at the 2026 Life Sciences Patent Network (LSPN) Spring Meeting in Boston, recognising our impact on the way life sciences and chemistry IP teams draft, prosecute, and analyse patents.

Artificial intelligence is no longer a theoretical issue in patent drafting, and many firms are already using AI-assisted workflows in some form. The harder question now isn’t whether to use AI, but how to talk to clients about using it.

Key insights

• Focus on better drafting quality, enforceability, and fewer avoidable downstream problems.
• Walk through data handling so confidentiality and retention protections are easy to trust.
• Explain inventorship stays human and the attorney remains responsible for every word.
• Keep the process clear and documented so expectations stay aligned from day one.

For some clients, AI usage signals efficiency and modernisation. For others, it raises immediate concerns about confidentiality, inventorship, and quality control. Those concerns are legitimate, so the key is to approach conversations about AI in a way that is structured, transparent, and grounded in professional responsibility.

In practice, the most effective discussions with clients will focus on outcomes rather than technology.

This article provides a structured framework to use when talking to clients about using AI, such as Solve Intelligence’s Patent Drafting Copilot^TM, in patent drafting.

The EPO and IP Australia are launching a new PCT pilot programme on 1 March 2026 which will allow Australian applicants to designate the EPO as their International Searching and Preliminary Examining Authorities (ISA and IPEA). 

Given the EPO’s rigorous approach to clarity and support requirements, for this pilot programme to succeed, Australian applicants and patent practitioners will have to adapt to draft international applications with EPO-specific requirements in mind.

The launch of this pilot programme will add a new layer of complexity — (and opportunity) for patent practitioners. In a landscape where jurisdictional nuance can shape international search and examination outcomes, AI‑augmented tools such as Solve Intelligence's Patent Copilot^TM are becoming increasingly valuable.