Crossing the Pond Twice: A New Playbook for Patent Diligence Between the US and Europe

Co-authored by: Matt Grady, Shareholder, Wolf Greenfield & Emer Bollinghaus, Partner, Kilburn & Strode

This article draws on the diligence framework set out in IP Diligence in the Age of AI: Why Standard Review Is No Longer Enough and AI Patents: How They Pass the Modern Diligence Test, together with discussion of European patent protection from Kilburn & Strode’s webinar, “Before You Launch in Europe: The IP Checklist.”

The winds of IP change are blowing across the Atlantic. US courts are applying a new level of scrutiny to patent subject matter, particularly to AI. The European patent system has a faster, further-reaching court than it has ever had before. Both shifts create new opportunities and new risks, which land on patentees and product makers alike. In practice this affects most companies, since a business that holds patents is almost always also selling a product. The instinct in diligence has long been to run one clearance exercise, build one portfolio strategy, and assume the result travels. That instinct no longer holds. The single-jurisdiction playbook has to be replaced with one that looks from the US to Europe and back again, twice: once at the patent protection in play, both the company’s own and any third party’s, and once at the product itself. This article explains why that second look matters so much more than it used to, and what doing it well actually requires.

Patent Protection Is Changing on Both Sides of the Pond

For years, the working assumption among many practitioners was that Europe was the slower, gentler jurisdiction for patent enforcement, while the US was the more forgiving one for getting software and AI patents granted in the first place. Neither assumption is safe any longer.

In Europe, the Unified Patent Court (UPC) has given patentees a tool with a reach and a speed that the old system of country-by-country litigation never offered. The UPC’s Düsseldorf Local Division’s 2026 decision in Hologic, Inc. v. Siemens Healthineers is a useful illustration. The court found that Siemens’ MAMMOMAT B.brilliant mammography system infringed Hologic’s patent on flying-focal-spot technology, rejected Siemens’ validity challenge in full, and granted a permanent injunction across Germany, France, and the Netherlands, together with orders for recall and destruction of systems already in the field, within a two-month time window, before hefty financial penalties would come into play. Siemens had argued that recall was disproportionate, including as applied to units already installed at universities. The court was not persuaded. A single first-instance decision, in one division, removed a competing product from three markets and required it to be pulled back out of the field. That is a materially different risk profile from the national litigation patentees faced before the UPC existed.

At the same time, the US has become a harder jurisdiction in which to obtain, and keep, a granted patent in certain technology fields. The Federal Circuit’s decision in Recentive Analytics, Inc. v. Fox Corp. found patent claims directed to using machine learning to optimize television schedules ineligible under Section 101, because the claims described no specific improvement to the machine learning technology itself. In re Brian McFadden reached a similar result at the USPTO, rejecting broadly drafted AI claims that never explained what made the underlying technology different from conventional computing. Subject matter eligibility scrutiny of this kind is no longer an edge case. It is a central feature of US patent practice for software and AI inventions, and it is producing narrower claims, harder validity fights, and patents that do less work as negotiating or defensive tools than their owners expected.

Put the two trends together and the effect on risk is mixed rather than uniform. A would-be infringer in the US, in a technology field where subject matter eligibility is being applied aggressively, may reasonably feel emboldened. The patents standing in their way are more likely to be narrow, more likely to be vulnerable to a validity challenge, and less useful to their owner as a basis for a confident assertion. On the other side of the Atlantic, the calculus runs the other way. A patentee with a European patent right now has a faster, more powerful court behind it, including the prospect of a far-reaching preliminary injunction obtained relatively quickly. A product maker who might once have treated a European patent as a background risk, worth noting but not losing sleep over, now has real reason to think twice before stepping into patented European territory.

Product Variation Is the Overlooked Variable

The second look diligence needs to take is not about the patents at all. It is about the product. US and European versions of the same product are routinely different, driven by differences in utilities infrastructure, local culture, and consumer demand, and that reality has two consequences that are easy to miss.

The first is on the clearance side. A freedom-to-operate opinion is only as good as the product it was written against. Even where the patent claims at issue are identical in the US and Europe, which they frequently are not, the product details behind a US clearance opinion may not hold for the European version of the same product. A non-infringement conclusion built on a specific component, material, or configuration in the US product does not automatically transfer to a European product that achieves the same function differently. Treating a US clearance result as proof of European freedom to operate, or the reverse, is accepting a conclusion that was never actually tested against the product now being sold.

The second consequence runs the other way, into the strength of the company’s own portfolio. A patent application drafted around a US product’s specific implementation may simply not read on the European version of that same product, even where the invention is conceptually identical. A company that assumes its patent protection automatically extends to cover its European product, without checking the claim language against the European product’s actual features, risks discovering that its own portfolio does not cover what it is actually selling in that market. That is wasteful in itself, and it weakens the portfolio in every direction at once: as a defensive shield, as a negotiating asset, and as a tool for stopping direct copies. Few things undercut a company’s credibility with investors or partners faster than a European competitor launching a close copy of its flagship product, unchallenged, because the European patent claims were drafted for a product that is not quite the one on the market.

How Do You Succeed? The Devil Is in the Detail.

None of this means European expansion or AI patenting should be treated as uniquely hazardous. It means that shortcuts carry more risk today than they used to, and that a few details now deserve attention earlier in the process rather than being left for litigation to expose.

On the US side, there are established ways to draft and amend patent applications to address subject matter eligibility scrutiny directly by tying claims to a specific technical mechanism, a measurable performance improvement, or a concrete architectural feature, rather than describing a generic model applied to a business problem. Diligence on a pending or recently granted US portfolio should confirm that this work has actually been done, claim by claim, rather than assumed from the fact that the application issued.

On the European side, the EPO applies its own version of this inquiry through the problem-solution approach, asking whether a claim identifies a genuine technical effect and solves an objective technical problem. A claim built to satisfy US eligibility case law is not automatically built to satisfy EPO practice, and the reverse is equally true, so a global portfolio review has to run both tests rather than treating success under one as a proxy for the other.

Coverage itself also needs to be checked directly rather than assumed. An EP patent is not automatically a unitary patent, and even where unitary effect has been requested, a unitary patent does not cover every EPC contracting state, only the EU member states that have joined the Unitary Patent system. A patent that looks commercially threatening on paper may have no effect at all in the specific countries where a company actually operates, or it may have broader reach than expected if validated as a unitary right. The UPC itself is not yet in force across every EU member state, and plenty of European countries sit outside the EU altogether, so a UPC-wide answer is never the end of the analysis. Confirming validation and court coverage, market by market, is no longer a detail that can be left until a dispute actually arises.

A Global Version of the Foundational Test

The foundational questions that determine whether a patent portfolio actually supports a transaction have not changed. What has changed is that each one now has to be asked on both sides of the Atlantic, because a portfolio and a product can pass the test in one jurisdiction and fail it in the other.

Does the portfolio match the product in every market where the product is actually sold? This means checking claim language against the real European product, not just the US original, and confirming that the company has searched the surrounding European patents with the same rigor applied in the US, including national rights that a US-centric search would never surface.

Is the portfolio in good order, jurisdiction by jurisdiction? Ownership, inventorship, and assignment records need to hold up everywhere a patent family has been filed or validated, and validation and maintenance status should be confirmed country by country and, where relevant, under the Unitary Patent system, rather than assumed from the fact that an EP patent was granted.

Do the claims hold up under both the US and European eligibility frameworks? A patent that survives Alice and Section 101 scrutiny because it describes a specific technical mechanism is far better positioned to satisfy the EPO’s technical-contribution requirement too, but the two tests are not interchangeable, and a portfolio built on generic “AI applied to X” claims is exposed on both sides.

Has the company assessed its offensive and defensive position under the current European system? Doing so means a considered view on unitary patent strategy, UPC litigation exposure given decisions like Hologic v. Siemens, and the national and EPO-level tools, such as oppositions and third-party observations, that remain available alongside it.

A portfolio that answers all four questions well on one side of the Atlantic but has not been checked on the other is not just half protected. It is protected in the jurisdiction nobody is likely to sue in next, and exposed in the one where the real risk, and the real opportunity, now sits.

Related content: