← IP News

Attack a Standard, Lose the Patent: UPC Paris Revokes KeeeX’s C2PA Patent on a Counterclaim

Attack a Standard, Lose the Patent: UPC Paris Revokes KeeeX's C2PA Patent on a Counterclaim
Image: FrDr, CC BY-SA 4.0, via Wikimedia Commons ยท source

The Paris local division killed EP 2 949 070 for added matter and lack of novelty on the defendants' revocation counterclaim, ending a claim that started at EUR 5.6 billion. Three lessons for anyone asserting a patent against a published standard.

The Paris local division of the Unified Patent Court handed down its decision on the merits in KeeeX’s action against the developers of the C2PA content-provenance standard on 17 November 2026, and the patent did not survive it. According to JUVE Patent, the panel — presiding judge and judge-rapporteur Camille Lignères, legally qualified judges Carine Gillet and Peter Tochtermann, and technically qualified judge Alessandro Sanchini — found every contested claim of EP 2 949 070 invalid, either for added matter or for lack of novelty, and dismissed the infringement claims as a consequence. KeeeX, a Marseille technology company, had sued Adobe, OpenAI, Truepic, the Joint Development Foundation and the Coalition for Content Provenance and Authenticity at the Paris division in June 2025, arguing that the C2PA standard reproduced the teaching of its patent. The defendants answered with revocation counterclaims, which are the claims that decided the case (JUVE lists the action numbers as UPC_CFI_530/2025, UPC-CFI-0000003/2026, UPC-CFI-0002285/2025, UPC-CFI-0002281/2025 and UPC-CFI-0002274/2025).

EP 2 949 070, titled by ip fray as a verification process for the integrity of a block of numerical data, covered inserting a digital fingerprint into a file and later recalculating it to confirm the file’s integrity and authenticity. The money at stake had already shrunk considerably before the ruling: JUVE reports that KeeeX originally sought EUR 5.6 billion in material damages plus EUR 1 billion in non-material damages, then reduced its request to EUR 100 million and EUR 20 million as interim damages. The court never reached the damages question. KeeeX bears all legal costs, with an interim cost provision set at EUR 300,000 split evenly at EUR 75,000 across the four defendant groups — an amount the panel moderated in view of KeeeX’s status as a small enterprise and the overlap between the defendants’ positions. An appeal is available. Separately, Adobe’s nullity action against the German part of the same patent remains pending at the German Federal Patent Court as 6 Ni 61/25.

The counterclaim is the main event, not the defence

The structural point behind this outcome is one that still catches claimants who are used to national practice. At the UPC a revocation counterclaim filed in answer to an infringement action is decided by the same panel, on the same timetable, in the same decision. There is no bifurcation to wait out and no separate nullity forum whose slower clock can be used as leverage. That means the realistic question before filing is not whether your opponent can show non-infringement within a year; it is whether your claims as granted can withstand a well-funded, fully briefed validity attack within the same year. Both grounds that landed here — added matter and lack of novelty — are the documentary grounds. They are proved from the application file and from prior art, not from expert argument about what a skilled person would have thought, and that makes them the cheapest grounds for a defendant to run and the hardest for a patentee to talk around.

Added matter in particular is a vulnerability created during prosecution, often years before anyone contemplates litigation. A feature lifted out of a worked example and written into a claim without the rest of the example reads perfectly well at the time and is fatal later. If you intend to assert a European patent at the UPC, the amendment history of the family is a litigation asset or a litigation liability, and it is knowable before you file suit. A clear-eyed invalidity assessment of your own patent — run as the other side would run it, against the application as filed and against the full prior-art field — belongs in the pre-filing budget of any assertion where the defendant will plainly counterclaim. For a target the size of a standards consortium, that is every assertion.

Suing the specification means suing a coordinated defence

Where the asserted reading of the claims is that a published standard infringes, the defendant is effectively the specification, and every implementer has the same defence and the same interest in funding it. The cost order in this case makes the economics explicit: the panel reduced the interim provision partly because the defendants’ positions overlapped, which is another way of saying a joint defence is efficient and the court can see that it is. A patentee who names five parties over one standard has not multiplied its leverage five times; it has consolidated the opposition and handed it the savings. The inverse holds for implementers. If your product reads on a published standard that someone has started asserting against, your exposure is shared and so, usually, is the best prior art — which is an argument for mapping the risk through an organised freedom-to-operate review of the standard itself rather than waiting to be served individually.

Do not plan on reaching non-UPC states by default

The third change long predates this week’s ruling and is the one most likely to affect filing and enforcement planning. The UPC Court of Appeal held on 13 March 2026 that the Paris local division had no jurisdiction over the Swiss, Spanish, British, Irish, Norwegian and Polish parts of the patent. As LaVoix summarises the order, the appellate panel read Article 7(2) of Regulation 1215/2012 as confining jurisdiction to damage occurring within the court’s own territory, and found that KeeeX had not substantiated infringing acts outside France — it asserted the harm was global without evidencing where it occurred. The practical rule the order sets is procedural rather than doctrinal: a claimant who wants extended, cross-border reach has to plead and evidence the basis for it in the statement of claim, not raise it once the defendants object.

For portfolio planning, that converts coverage in non-UPC European states back into something you handle deliberately. Spain, Poland, the United Kingdom, Switzerland, Norway and Ireland are either outside the UPC Agreement or outside the EU jurisdiction regime that would carry a UPC order into them, and a single UPC action is not a reliable substitute for enforcing there. If those markets matter commercially, the validation and enforcement budget has to treat them as separate proceedings. And if the plan is to reach them through a UPC action anyway, the extraterritorial case has to be built into the first brief, with evidence of acts in each territory, rather than left as an assumption.

One last number worth keeping. From filing in June 2025 to a merits decision revoking the patent took roughly seventeen months, and that included a jurisdictional appeal to the Court of Appeal and a run of procedural skirmishes over interim damages and claim amendments. That is the planning horizon the UPC now offers both sides — fast enough that a weak patent gets tested before the commercial dispute matures, and fast enough that an unprepared claimant does not get the years of settlement pressure the forum was once expected to generate.

Sources

Reported by PerspireIP from the sources above. This is general information about developments in intellectual property law, not legal advice on any specific matter.