Should Abandoned Applications Be Presumed Enabling? Supreme Court Asks for Response

by Dennis Crouch

The Supreme Court has called for a response in Agilent Technologies, Inc. v. Synthego Corp., No. 25-570, a petition that challenges foundational assumptions about the use of prior art references. The response is due in January 2026. The basic question in the case is whether printed publications asserted as anticipatory prior art should be presumed enabling, and whether the Federal Circuit’s holding in Rasmusson v. SmithKline Beecham Corp., 413 F.3d 1318 (Fed. Cir. 2005), that “proof of efficacy is not required” for a reference to anticipate, should be vacated or narrowed.

In this case, the allegedly anticipatory reference is an abandoned patent application filled with prophetic examples describing modifications that, according to Agilent, nobody has ever shown to work. The petition asks the Court to reconsider whether modern patent law has drifted too far from the principle articulated in Seymour v. Osborne, 78 U.S. (11 Wall.) 516 (1870), that a prior art reference must actually “communicate” the invention to the public in a meaningful sense to defeat a patent.

The Supreme Court’s call for response (CFR) carries some procedural significance. The Court receives thousands of certiorari petitions each year and grants fewer than 80. Most petitions are denied without any response from the opposing party. When the Court requests a response, it signals that at least one Justice believes the questions warrant fuller consideration. This does not mean certiorari is likely; most CFR cases still result in denial. But it elevates the petition above the thousands that receive no engagement whatsoever. For a patent case raising questions about anticipation doctrine, this level of interest is worth noting. The enablement requirement for anticipatory prior art is not a question the Court has addressed directly in the modern era, and the petition frames the issue as one of growing practical importance as the nature of “printed publications” continues to shift.

The Enablement Presumption for Prior Art

When a patent challenger argues that a prior art reference anticipates a patent claim, courts apply a presumption that the reference is enabling. This means the challenger need not affirmatively prove the reference teaches a person of ordinary skill how to make and use the disclosed invention. Instead, the patentee bears the burden of proving the reference is not enabling. The rationale traces to an assumption that published technical documents generally reflect real knowledge. But this presumption can be rebutted by showing the reference contains inoperable embodiments or would require undue experimentation to practice.

The underlying dispute involves Agilent’s patents on modified guide RNAs used in CRISPR-Cas9 gene editing. Synthego challenged the patents before the PTAB, asserting anticipation based on a Pioneer Hybrid patent application. That application disclosed various guide polynucleotide modifications, including sequences presented in what the parties call “Table 8.” The catch is that Pioneer Hybrid was abandoned, its examples were entirely prophetic rather than actually performed, and Agilent contends the vast majority of what it disclosed simply does not work. The Federal Circuit heard oral argument in the appeal from the PTAB’s anticipation finding, and that case remains pending. But the Supreme Court petition challenges the doctrinal framework itself, regardless of how the Federal Circuit resolves the specific factual disputes.

The oral argument before the Federal Circuit, featuring Professor Mark Lemley for the patentee Agilent and Edward Reines for Synthego, illuminates the doctrinal stakes. Lemley framed two distinct but related arguments. First, that Pioneer Hybrid never actually disclosed functionality sufficient to put a person of ordinary skill in the art in possession of working guide RNAs. Second, even if some disclosure existed, the reference failed to enable because it “threw everything at the wall” without teaching which of the “quadrillion quadrillion possibilities” would actually function. Judge Prost pressed Lemley on whether these were anything more than substantial evidence arguments, asking whether the Board’s detailed factual findings should receive deference. Lemley responded that the legal question was whether a reference that says “try anything” can satisfy enablement when a person of skill would have to conduct extensive trial-and-error experimentation to identify operable embodiments.

In my mind, the “throw everything at the wall” problem should receive a special spotlight as AI-generated technical content proliferates. Large language models have been producing vast quantities of plausible-sounding scientific disclosures. These are synthetic publications that read as authoritative but are untethered from actual experimentation — and often from reality itself. Some of these AI-generated predictions will, by combinatorial luck, accurately describe functional embodiments. Yet nothing in their presentation distinguishes the prophetic hits from the prophetic misses; all carry the same veneer of technical respectability. Current anticipation doctrine, which presumes enabling disclosure and places the burden on patentees to prove otherwise, seems ill-equipped for a landscape where machine-generated prior art can flood the technical literature with untested possibilities. Requiring meaningful proof that a reference actually enables what it purports to disclose is one doctrinal lever for ensuring that anticipation remains grounded in genuine contributions to public knowledge rather than algorithmic speculation.

The case is also centered on the Supreme Court’s recent decision in Amgen Inc. v. Sanofi, 598 U.S. 594 (2023). In Amgen, the Supreme Court addressed enablement as a validity requirement for a patentee’s own claims, holding that a patent claiming an entire genus must enable the full scope of what it claims.  But, the appellate court in Agilent found that Amgen has nothing to do with whether prior art is enabling.

Prophetic Examples in Patent Applications

Patent applications often include "prophetic examples" that describe experiments the applicant has not actually performed. These are written in future or conditional tense ("the compound would be expected to exhibit activity") rather than past tense describing completed work. Prophetic examples are legally permissible but must be reasonably based on the applicant's knowledge. When a reference containing only prophetic examples is asserted as prior art, questions arise about whether it 
truly enables what it describes or merely speculates about possibilities the applicant never reduced to practice.

The broader context makes this case timely. As I discussed in connection with the Lynk Labs v. Samsung petition, the presumption that published materials are enabling reflects assumptions about publication that may no longer hold. See Dennis Crouch, Will the Supreme Court Cut the Gordian Knot of Prior Art Enablement?, Patently-O (Nov. 2025). Historically, “printed publications” involved gatekeeping. If something appeared in print, there was reason to believe it reflected genuine knowledge rather than speculation. The current landscape looks different. Patent applications publish automatically 18 months after filing regardless of technical merit. Preprint servers host unreviewed manuscripts. The proliferation of AI-generated content means that more and more online text is divorced from any underlying physical reality or experimental verification. A document can describe technical procedures in confident, detailed language without any human having performed (or even reviewed) them.

This evolution calls into question whether a blanket presumption of enablement remains appropriate. The presumption makes more sense for issued patents being asserted as prior art – they have been examined for enablement and deemed worthy.  It makes less sense when the category of “printed publications” includes abandoned applications filled with prophetic examples, AI-generated technical documents, and preprints that may never survive peer review. The Pioneer Hybrid reference at issue here is an abandoned patent application. Its examples are prophetic. According to Lemley at oral argument, “the reason it was abandoned, the reason they’re still trying 12 years later with a very different set of patent claims to get something, is their proposal, their approach, didn’t work.” If that characterization is accurate, the reference describes things its own applicants could not achieve. Yet under current doctrine, it is presumed to enable what it describes.

Judge Reyna’s question at oral argument touched on this concern. He asked Reines to address the impact of Pioneer Hybrid’s abandonment. Reines responded that abandonment had “no impact” because the presumption of enablement “gets us 90% of where we need to go.” This is doctrinally accurate under Federal Circuit precedent. But it also illustrates the petition’s point. The presumption does most of the work. A challenger can point to a reference that may describe inoperable subject matter, invoke the presumption, and force the patentee to prove a negative.

There is also an important distinction between patent prosecution and contested post-issuance proceedings that the current doctrine obscures. During ex parte prosecution, the presumption of enablement makes more practical sense because the examiner lacks the resources and institutional capacity to prove non-enablement. An examiner reviewing dozens of applications cannot conduct experiments, retain experts, or depose inventors. The presumption allows examination to function. But this administrative convenience comes with a built-in safety valve: the applicant can easily amend claims. Prosecution is iterative and more forgiving; the cost of an initial erroneous rejection is delay, not destruction.

In IPR or litigation, the challenger typically has full capacity to develop evidence of enablement. The patentee, meanwhile, loses the ability to amend. The claims that issued are the claims that must be defended. And critically, those claims now carry a statutory presumption of validity under 35 U.S.C. § 282. There is something doctrinally awkward about simultaneously presuming that an issued patent is valid and that the prior art reference asserted against it is enabling. Both presumptions cannot be doing honest work.

As we do with many other patent doctrines, it is reasonable to consider whether the prior art enablement presumption should carry the same weight across these contexts. A rule designed to facilitate efficient examination by under-resourced examiners need not govern adversarial proceedings where sophisticated parties can develop actual evidence about whether a reference defeats an issued patent.