FDA has undergone fundamental changes under the Trump administration. The agency is smaller, but is it smarter? Commissioner Makary has embraced AI and appears to be looking for new ways to abbreviate clinical development for innovative cures. Meanwhile, the courts are continuing to shape Orange Book patent listing practices, and the Supreme Court is set to opine on the implications of labeling carve-outs. We take stock of these developments and consider five areas in which we expect further consolidation and change.
1. Skinny labeling
By statute, generic products must have the same labeling as the innovator drug they reference (the reference listed drug, or RLD). There is a critical exception: a generic may omit or “carve out” a patented method of use and seek approval for “skinny labeling” that seeks to avoid infringing the patent. FDA will approve the generic with this skinny labeling provided the product as approved is no less safe or effective for the remaining, non-protected conditions of use.
There are a number of longstanding problems with this approach. First, FDA assigns an “A” therapeutic equivalence rating to all approved generic drugs, which in most states results in the generic being automatically substituted for the brand-name drug at the pharmacy, regardless of whether the prescribed use has been carved out of the generic labeling. Thus, while skinny labeling may be nominally non-infringing, the product will, as a certainty, be dispensed for the patent-protected method of use. Second, the carve-out process is not precise as FDA does not interpret the scope of a patent. Rather, the agency relies on a short “use code” statement from the innovator sponsor describing what the patent covers when determining whether proposed skinny labeling has adequately carved out the protected use. In some cases, the generic sponsor may carve too narrowly for patent infringement purposes, and a residue of information about the protected method of use may remain in the labeling. Third, assuming a protected use is successfully carved out of the generic labeling, other promotional statements made by the generic sponsor may encourage use of the product “off label” for the protected use, or suggest that the generic product is approved for more uses than what its labeling includes.
For these reasons, a generic with skinny labeling will routinely be used for the patent protected condition of use, and the generic manufacturer can be subject to liability for induced patent infringement. To establish a claim of induced infringement, plaintiffs must allege 1) that the defendant “actively” induced infringement, 2) that the defendant knew its acts would induce infringement, and 3) that the induced use constitutes direct patent infringement.
There has been considerable controversy in recent years relating to the intersection of labeling carve-outs and patent infringement. The first major test for induced infringement came in 2021 with a landmark Federal Circuit decision in GSK v. Teva. The district court in that case entered a directed verdict of non-infringement, overturning a jury verdict of infringement. The Federal Circuit in turn reversed, holding that a jury could reasonably find statements made in both the generic label and in the press as inducing infringement. The Supreme Court declined to weigh in on the matter.
In 2024, the Federal Circuit again addressed the issue of induced infringement in Hikma v. Amarin. There, the court held that—although the generic labeling, on its own, was insufficient to give rise to a claim of induced infringement—when considered with public statements made by the generic sponsor, the case could proceed to trial. The generic sponsor appealed, and this time the Supreme Court granted certiorari on January 16, 2026, as we wrote about here.
The case will be heard this spring, and the Supreme Court will issue its opinion in late summer or early fall. Some have speculated that the Court may craft a safe harbor wherein a generic with skinny labeling is broadly insulated from a claim or induced infringement; however, that seems unlikely. The outcome should clarify the conditions under which generic manufacturers may be held liable for induced infringement.
2. Biosimilars and interchangeability
In 2010, the Biologics Price Competition and Innovation Act (BPCIA) was passed as part of the Affordable Care Act. The BPCIA created an abbreviated approval pathway for biosimilars, or biological drugs that are highly similar to a reference product without any clinical meaningful differences in safety, purity, or potency.
At the time of its enactment, the BPCIA reflected a hard-fought compromise regarding automatic substitution of biosimilars by pharmacies. The substitution question for biosimilars is different from the typical substitution between brand-name and generic small-molecule drugs because unlike those drugs, biosimilars are not identical to their reference products. The enacted law reflects a compromise between access and patient safety by creating two types of biosimilars—biosimilars and interchangeable biosimilars—which are supported by different data packages. This two-tiered system reflects the concern that, without additional data, usually in the form of a switching study, automatic substitution of biosimilars would involve greater risk to patient safety than the typical substitution used for small-molecule generics. Under this system, only interchangeable biosimilars, or “interchangeables,” typically supported by a switching study, are automatically substituted by pharmacies (subject to state law). General biosimilars have a lower burden because they do not require any type of switching study. However, these biosimilars have not been historically substitutable by pharmacies, a reflection of their lower burden.
In the last two years, FDA has begun to collapse the difference between these two types of biosimilars, i.e., to reduce the bar for a finding of interchangeability. For instance, in 2024, FDA announced it would no longer apply a general requirement for a switching study for interchangeables. See Draft Guidance for Industry: Considerations for Demonstrating Interchangeability with a Reference Product: Update (Jun. 2024). In October 2025, the agency took an additional leap, issuing a revised draft guidance stating it would no longer routinely require comparative efficacy studies to demonstrate baseline biosimilarity. See Draft Guidance for Industry: Scientific Considerations in Demonstrating Biosimilarity to a Reference Product: Updated Recommendations for Assessing the Need for Comparative Efficacy Studies (Oct. 2025).
This shift in policy narrows the distinction between biosimilars and interchangeables, and substantially reduces the data required to demonstrate baseline biosimilarity.
With the current administration’s interest in lowering drug costs and increasing affordability, it is likely to continue considering ways to encourage biosimilar entry. See Fact Sheet: Bringing Lower-Cost Biosimilar Drugs to American Patients (Oct. 2025). This will be good news for biosimilar sponsors.
3. Patent listing
An integral part of the Hatch-Waxman framework is that innovator sponsors must submit information to FDA on patents that claim the drug substance (active ingredient), drug product (formulation and composition), or method of using the approved drug. The information on these patents is listed in the Orange Book. Determining what patents must be listed is of great importance. ANDA or 505(b)(2) NDA sponsors referencing the innovator drug and seeking to come to market before a patent expires will first need to certify that the patent is “invalid or will not be infringed.” This is known as a Paragraph IV certification. Alternatively, the applicant can wait for the patent to expire or, for method of use patents, seek to “carve out” the protected use. If the applicant submits a Paragraph IV certification challenging the listed patent as invalid or not infringed and the patent holder sues for infringement within 45 days of receiving the Paragraph IV certification, the ANDA or 505(b)(2) NDA will be subject to a 30-month stay of approval.
Despite the significant consequences of listing a patent in the Orange Book, it is not always clear when a sponsor should do so. This is particularly true in the case of combination products. For example, in in re Lantus Direct Purchaser Antitrust Litigation, the First Circuit was asked whether Sanofi wrongly listed patents for the drive mechanism of the drug delivery system for its product. Sanofi argued that its pre-filled drug delivery system is a “dosage form,” and that the term “drug product,” as defined by FDA, includes dosage forms. In a 2020 decision, the court rejected that argument. It explained that, although a filled drug delivery system may very well be a dosage form (patents for which would ordinarily be listable in the Orange Book) the patents for the drive system, alone, are not listable.
The issue of whether delivery system patents should be listed in the Orange Book was again raised in Teva v. Amneal. Here, Teva listed patents claiming improvements to the device component of the product (e.g., the inhaler dose counter and the inhaler canister). Teva explained that it read the Food, Drug, and Cosmetic Act as requiring the listing of any invention that was part of its approved product. In December 2024, the Federal Circuit held the patents were improperly listed. In doing so, the court explained that not all patents claiming components of a drug are listable. Rather, “to qualify for listing, a patent must claim at least what made the product approvable as a drug in the first place—its active ingredient.”
The Federal Trade Commission (FTC) under President Biden engaged in aggressive and highly public action against sponsors who had listed device patents in the Orange Book. The current administration has been equally forceful on this issue. This is a longstanding issue at the agency. On several occasions, FDA has been asked whether drug delivery system patents that do not reference the approved active ingredient or formulation should be listed. In all instances, the agency has declined to provide additional guidance, stating that it has a “purely ministerial” role when it comes to listing patents in the Orange Book.
In a 2022 report to Congress, FDA noted that there is no consensus answer on whether patents like those in Lantus and Teva should be listed, but committed to forming a multidisciplinary working group to “evaluate whether additional clarity is needed.” In January 2024, CDER included on its regulatory agenda a guidance titled Submission of Patent Information for Listing in the Orange Book. Although no such guidance has been issued, it indicates that FDA may be developing a position on whether drug delivery system patents should be listed. Underscoring all this, a January 2026 report from the Congressional Research Service suggested that Congress may “consider whether to impose more responsibilities on FDA, FTC, or the courts, or whether to expand current procedures for challenging Orange Book patents before FDA or in court.”
4. Commissioner’s national priority vouchers
In June 2025, FDA unveiled the Commissioner’s National Priority Voucher (CNPV) program, as we wrote about here and here. Under the program, vouchers will be issued to align with specific US national health priorities, namely whether they (1) address a US public health crisis, (2) deliver more innovative cures for the American people, (3) address a large unmet medical need, (4) onshore drug development and manufacturing to advance the health interests of Americans and strengthen US supply chain resiliency, or (5) increase affordability. As of January 2026, there have been 16 CNPV recipients, and in December 2025, the first CNPV product, Augmentin XR, was approved.
To be awarded a CNPV, a company must submit to FDA a short statement of Interest (350 words or fewer), highlighting the national health priorities the product will address. Each CNPV will come with the promise of enhanced communication from the agency and an expedited review window of 1-2 months, down from the usual review goal of 10-12 months under the Prescription Drug User Fee Act. Chemistry, manufacturing, and controls information must be submitted at least 60 days before the complete application is submitted if the drug is to be manufactured domestically, and at least 120 days before if foreign inspections will be required.
FDA reserves the right to extend its review window “if the data or application components submitted are insufficient or incomplete, if the results of pivotal trial(s) are ambiguous, or if the review is particularly complex.” Nonetheless, those granted a CNPV have the opportunity to see their products approved and available to patients in record time. The agency is adopting a tumor-board style of review for CNPV products. A cross-disciplinary team, composed of scientists and physicians, convene for one day to determine whether the product should be approved. To facilitate the fast-moving nature of this process, members will have the opportunity to review the application package prior to a full meeting of the board.
The program is not without controversy, however. It was created without any formal process, and it has raised concerns about transparency and governance. FDA has not published detailed, objective criteria for how “national priorities” are defined or how voucher recipients are selected. Key insights will come from how the program evolves in 2026.
5. Artificial intelligence
Over the past year, FDA has embraced the use of AI. In December 2025, the agency announced it would be implementing agentic AI capabilities for all agency employees. Agentic AI refers to advanced artificial intelligence systems which are designed to reach goals by planning, reasoning, and executing multi-step actions. At this time, use of the tool remains optional and voluntary. This follows the May 2025 agency-wide rollout of the language-learning model, Elsa, which is reportedly now used by more than 70% of staff.
The agency’s attempts to incorporate AI have walked the fine line between using as much data as possible (to better train the model) and ensuring data submitted to the agency is appropriately safeguarded. According to FDA, while Elsa and the newer agentic capability are intended for use with product review, current models are not trained on data submitted by regulated industry. But as AI tools continue to improve and become increasingly incorporated into both FDA’s internal mechanisms and into the products that the agency reviews, there are bound to be controversies. The use of agentic AI in premarket review is new. There are statutory limits to FDA’s authority to access sponsor-submitted data to inform its review of a subsequent marketing application. It remains unclear how the technology will police this line.
