The IP Brief - July 2026

In the EU, the European Parliament adopted its (equally pro-copyright) resolution on copyright and generative artificial intelligence on 10 March (EP Resolution). This was later followed by the European Commission launching a Call for Evidence to support its review of the 2019 Directive on Copyright in the Digital Single Market (DSM Directive) (which contains, amongst other things, the EU’s copyright exceptions for text and data mining (TDM) and to inform potential targeted Commission proposals to change the EU’s copyright framework, including in the context of AI.

We summarise some of the key points emerging from these publications below.

UK

One of the core debates within the UK (and globally) over the last few years has been about how to balance the rights of AI developers and copyright holders when training generative AI. That debate encompasses a number of material topics including the availability of copyright exceptions to AI developers, transparency over data sources, licensing and remuneration, technical tools and standards, and the treatment of models trained abroad.

Given the UK government’s current approach (discussed further below) of largely deferring policy decisions until a later date, as well as the limited nature of the issues ultimately argued (and pending before the Court of Appeal) in Getty Images v Stability AI, we suspect we might now start to see further litigation to clarify the legal position in the UK from those who had been waiting for the legislature to intervene.

TDM exception

Of these, the one that has attracted the most attention to date in the UK has been whether a broader copyright exception for TDM should be introduced. After a failed attempt in 2022 to introduce an exception which would have allowed TDM for any purpose with no ability for rights holders to opt out or contract out, the government stated a preference in its December 2024 consultation on AI and copyright (Consultation), to introduce a slightly narrower TDM exception that would have permitted TDM for any purpose but only where rightsholders had not reserved their rights and opted their content out (largely mirroring the current position in the EU).

Following significant pushback from those in the creative industries, the government has now confirmed that this is no longer its preferred option. It has not, however, put forward an alternative preference at this stage, opting instead to gather further evidence. For now, therefore, the status quo will continue, with the UK only having a very narrow TDM exception for non-commercial research purposes.

It will be interesting to see whether this retreat has any impact on the developing licensing market. In the meantime, it looks like we’re back in a holding pattern, with both rightsholders and AI developers in a position of uncertainty as to what the future direction of travel is likely to be.

Transparency

As things stand, there is no requirement in the UK for AI developers to publish details of the works used to train their generative AI models or how those works were obtained.

The CDC Report called on the government to set best practice for transparency reporting on the use of data in AI training and to place transparency obligations for large AI model providers on a statutory footing.

Responses to the Consultation indicated that there is broad consensus for greater transparency, but, unsurprisingly, there are differing views on approach - particularly on the appropriate level of granularity and how information should be disclosed.

With that in mind, the government has said that it intends to continue monitoring the effect of transparency rules in other jurisdictions, such as the EU and California, and will publish a review of mechanisms available for creators to control the use of their works online, including details of best practice on input transparency, in due course (no timeframe provided).

Licensing

With respect to licensing, the CDC Report urged the government to set out a clear expectation that commercial AI developers operating in the UK must obtain appropriate licences to use copyright works for training generative AI, and that appropriate mechanisms should be in place to ensure fair remuneration for individual rightsholders.

The government’s response did not set out the exact statement requested by the CDC, but did note that “AI developers must seek permission and acquire licences for the use of copyright works, unless they can show an existing copyright exception applies” and that “licensing is important to ensure creators are paid when their work is used”. At the same time, it reiterated that the government does not, generally, intend to intervene in the licensing market at this stage, although it acknowledged that more might need to be done to support individual and smaller creative rightsholders. It therefore intends to launch a working group to explore whether there is a role for it to play in providing such support.

Technical tools / standards

On the topic of technical tools and standards, the CDC Report recommended that the government develop interoperable standards for labelling AI-generated content (as well as technical solutions to reserve rights and show content provenance) and consider enshrining labelling requirements for wholly AI-generated content in legislation.

In response, the government has said that it will: (i) establish a taskforce on best practices for AI-generated content labelling, with an interim report due this autumn; and (ii) take forward work with experts and stakeholders to support best practice and adoption of market-led tools and standards.

Treatment of models trained abroad

The Government Report considered whether UK copyright law should apply to models that have been trained overseas and then placed on the UK market. A related question was considered last year in the Getty Images v Stability AI case, where the English High Court was asked whether importing a trained generative AI model into the UK could amount to secondary infringement of copyright (more on this here and here).

The government noted the potential negative effect such extra-territorial reach could have, such as models not being made available in the UK, and the possible negative impacts on downstream developers of AI systems (which are more likely to be UK based than providers of the major generative AI models themselves). It concluded that this is an area of the law that should be considered properly by the English courts, based on the law as it stands. It also acknowledged that developments in other countries (particularly the EU and US) could have a significant impact on outcomes in the UK, with greater clarity expected in those jurisdictions over the coming year or two.

On this basis, the government has proposed not to amend UK copyright law in respect of models trained abroad for now, but will continue to monitor the situation and engage with other countries on the point.

Computer-generated works (CGWs)

The UK is one of the few countries in the world that provides copyright protection for CGWs, but there are questions around how those rules currently apply in a generative AI context. Acknowledging those issues, the government consulted on a number of potential reforms, with its preference being to remove the CGW provisions unless sufficient evidence emerged on the positive effects of protecting CGWs.

The Government Report noted that most respondents to the Consultation supported the removal of these provisions and the government has so far received limited evidence that the existing protections are being used. Whilst the government didn’t take the opportunity to make a final decision on this point, preferring to continue monitoring the use and impact of these provisions, the preference is still very much to remove these provisions unless strong evidence to the contrary arises.

Digital replicas

Unauthorised digital replicas, including those mimicking a person’s image or voice, are an area of growing concern in the UK (as elsewhere). As things stand, however, there are no specific personality or image rights in the UK and, whilst reliance on certain broader protections may be possible, the scope for such reliance is limited.

The Government Report noted that there was clear support from respondents to the Consultation for enhanced protections for image and voice, but no clear view on what form such protections might take or who may benefit from them. The government has therefore proposed to explore potential options to address the risks posed by digital replicas and confirmed that it will launch a new consultation focussing on this, this summer.

EU

Whilst the EU is notably further advanced than the UK in plotting its way forward on many of these issues (with the AI Act and its broader copyright exceptions for text and data mining already in place), it is clear that there is still plenty of room for debate and that the position in the EU is not settled.

The EP Resolution, for example, calls on the European Commission to reconsider the current approach, ultimately favouring a system based on licensing, remuneration and more granular transparency, over broad exceptions, with a number of recommendations and potential reforms proposed. These include:

• A call for clarification on the application of the DSM Directive in the context of training generative AI models.
• A call for an additional legal framework to clarify licensing rules for generative AI and to address potential infringements of current copyright law.
• A recommendation that the Commission assign responsibility to the EUIPO to support a sector-based, voluntary licensing process.
• A call for rightsholders to receive fair remuneration for the use of their works and consideration to be given to whether, and if so how, rightsholders should be paid for past uses of their works.
• A requirement for AI developers to disclose itemised lists identifying each item of copyright-protected content used for training (albeit with proposed solutions to protect trade secrets).
• The creation of a “rebuttable presumption” that copyright-protected content has been used for AI training where EU transparency obligations have not been met.
• A call for EU copyright law to apply to the use of European content for training any generative AI model or system that is made available on the EU market, regardless of where the model or system was trained, with any model or system that does not comply being barred from the EU market (building on the provisions already contained in Article 53(1)(c) of the EU AI Act, which we discussed in this briefing).
• A call for the Commission to investigate measures to protect individuals against the distribution and sharing of deepfakes.

Shortly after the EP Resolution was passed, on 13 May the European Commission launched a combined Call for Evidence. Through this, the Commission is seeking views and evidence on (amongst other things) the impact of the DSM Directive (which, importantly in this context, contains the EU’s existing TDM exceptions) as well as on, more broadly, the challenges relating to the application and enforcement of copyright in an AI context.

The Call for Evidence itself is relatively light on specific proposals. It notes that the Commission “is exploring ways to enhance the licensing and enforcement of copyright and related rights in the AI context, improve the conditions for creators’ remuneration, while making it easier for providers of generative AI to access copyright-protected content” and that “[p]olicy options may include improving rightsholders’ access to information on the use of their content where challenges or gaps are identified, facilitating licensing through mediation or arbitration or potential new measures to enable appropriate remuneration for creators, and strengthening protection of performers against AI-generated impersonifications”. However, at this stage, all options seem to be on the table, with the Commission seeking views to guide it towards a narrower set of proposals that it will then consult on.

This Call for Evidence may therefore provide stakeholders with an opportunity to influence the broader proposals on copyright and generative AI that the Commission will take forward into a stakeholder consultation.

In this latest instalment, the CJEU was asked, in essence, whether this type of sampling could fall within the scope of the pastiche exception set out in Article 5(3)(k) of the InfoSoc Directive (Directive 2001/29/EC).

More particularly, the CJEU was asked:

1. Whether the pastiche exception has a “catch all” nature which covers any artistic engagement with an existing work, including sampling, or whether the concept of pastiche is subject to limiting criteria, such as a requirement of humour, stylistic imitation or tribute?
2. In order for use to be “for the purpose of” pastiche, is it necessary to show that the user intended to use the protected work for that purpose (subjective test) or is it sufficient for the pastiche character to be recognisable to a person familiar with that work (objective test)?

Meaning of pastiche

In answering the first question, the CJEU sought to strike a fair balance between the rights of rights holders on the one hand and freedom of expression and of the arts, as well as the wider public interest, on the other.

Bearing that in mind, the CJEU held that the pastiche exception is not a “catch-all” and doesn’t cover every form of creative use of copyright-protected material.

Rather, pastiche covers creations which evoke one or more existing works, while being noticeably different from them, and which use (including through sampling) some of those works’ copyright-protected characteristic elements to engage with those works in an artistic or creative dialogue that is recognisable as such.

That artistic or creative dialogue may take different forms, such as overt stylistic imitation of the original work (concealed imitations or plagiarism, by contrast, are not covered), a tribute to the original work, or humorous or critical engagement with it (although humour or mockery is not required for a finding of pastiche).

Subjective or objective test?

In its answer to the second question, the CJEU held that there is no requirement to show that the user intended to use the existing work for the purposes of pastiche. Rather, whether use is “for the purpose” of pastiche is to be assessed objectively, with the test being met if the pastiche nature is recognisable to a person familiar with the work from which the elements have been taken.

Comment

It remains to be seen how the EU national courts will interpret and apply this decision in practice, but it nonetheless provides rare and helpful guidance on the scope of the pastiche exception which will be welcomed by rights holders and creatives alike.

For the music sector, the decision confirms that sampling can benefit from the pastiche exception in the right cases, but it does not give carte blanche to sampling so care still needs to be taken. Indeed, it will be interesting to see whether the German courts ultimately find the sample used in Nur mir to fall within the scope of the exception.

More broadly, we expect this decision will embolden more parties to raise the pastiche exception in future, not only in the context of music sampling, but also in other areas such as AI. In fact, pastiche has already been raised in an AI context in the UK in the Getty Images v Stability AI case, although ultimately the court did not need to rule on its application in that case after Getty dropped its claims for primary copyright infringement.

The Pelham II decision isn’t binding in the UK, but English courts are likely to have regard to it given (i) the English law version of the exception (which can be found in section 30A of the Copyright, Designs and Patents Act 1988) directly implements Article 5(3)(k) of the InfoSoc Directive, and (ii) the limited English case law on pastiche to date.

Building on earlier UPC case law, as well as the CJEU’s decision in BSH v Electrolux (see our summary here), the judgment clarifies the UPC’s approach to accepting jurisdiction over infringement claims concerning European patents designating non-UPC territories, based on the defendant’s domicile, and sets out a broader framework for exercising that jurisdiction.

Background

As discussed in our June 2025 edition of The IP Brief, over the last couple of years the CJEU and various divisions of the UPC have been considering the scope of their respective courts’ jurisdiction to hear cross-border patent infringement disputes.

Broadly speaking, it has been held that the relevant courts do have jurisdiction to hear cross-border infringement actions where the defendant is domiciled in the place of the court, even if validity of the underlying patent is challenged.

The UPC Court of Appeal’s recent decision in Kodak v Fujifilm takes this point further, confirming the position at appellate level for the UPC and giving more detailed guidance on how that jurisdiction should be exercised.

The underlying dispute

The dispute concerned one of Fujifilm’s European patents relating to lithographic printing plate precursors. Fujifilm alleged that Kodak’s Sonora Xtra-3 printing plates infringed its patent in both Germany and the UK and brought proceedings before the Mannheim Local Division of the UPC. Kodak denied infringement, counterclaimed for revocation of the patent and challenged the UPC’s jurisdiction over the UK designation.

At first instance, the Mannheim Local Division held the German designation valid and infringed. Three months later it found that it also had jurisdiction over the UK aspects of the infringement claim and, again, found infringement. It did not, however, accept jurisdiction to decide on the validity of the UK designation. Validity was only considered as a defence to the infringement claim, with that defence ultimately dismissed.

Kodak appealed, including on the UPC’s jurisdiction to rule on the UK designation.

Clarification of UPC’s long-arm jurisdiction

The UPC Court of Appeal confirmed the Mannheim Local Division’s finding that the UPC had jurisdiction to hear the infringement claim concerning the UK designation of the patent as the defendants were all domiciled in Germany.

In reaching that conclusion, the court dismissed Kodak’s arguments based on Article 34 of the UPC Agreement. Article 34 provides that UPC decisions concerning a European patent will cover the territory of the UPC contracting member states for which that patent has effect. Kodak argued that this meant that UPC decisions should only cover the UPC territories in which the patent has effect. The Court of Appeal disagreed, finding that Article 34 is not a territorial limit on the UPC’s jurisdiction. Rather, Article 34 simply clarifies that, as a rule (unless a more limited scope is requested), UPC decisions will cover all UPC territories where the patent has effect.

The court then went on to draw a distinction between accepting jurisdiction (based on the defendant’s domicile) and exercising that jurisdiction, noting that decisions to stay or dismiss a claim can only be made once jurisdiction has been accepted and that, in exercising jurisdiction, the UPC must apply international law principles of comity.

The framework for exercising jurisdiction

Perhaps most usefully, the judgment considered how jurisdiction over non-UPC designations of European patents should be exercised in three specific scenarios, drawing distinctions (where relevant) between cases where the non-UPC designations relate to (i) EU member states or signatories to the Lugano Convention (EU/LC EPs) and (ii) neither EU member states nor signatories to the Lugano Convention (non-EU/LC EPs).

Scenario 1 - Standalone revocation actions

For standalone revocation actions, the UPC should declare that it lacks jurisdiction regardless of whether the non-UPC designations in suit are EU/LC EPs or non-EU/LC EPs.

Scenario 2 - Infringement actions where the patent is held invalid in the UPC territories but would have been infringed if it were valid

In these types of infringement scenario, the patentee should be given an opportunity to withdraw the non-UPC infringement claim.

For EU/LC EPs:

• If the patentee does not withdraw the claim and no revocation action is already pending before the competent national court, the defendant should be given the opportunity to file such an action within an appropriate period of time.
• If a revocation action is pending or the defendant has lodged such an action, the UPC infringement proceedings with respect to the relevant designations should generally be stayed pending the national court’s final decision.
• If a revocation action is not pending and the defendant does not lodge a national revocation action within the allotted time, the UPC will assume that the patent is valid and will decide the infringement claim on that basis.

For non-EU/LC EPs, if the patentee does not withdraw the claim it will be dismissed, unless there are specific reasons not to do so (e.g. because the relevant claim is different and may be considered valid). 

Scenario 3 - Infringement actions where the patent is considered valid and infringed in the UPC territory

In this scenario, the UPC may, where appropriate, issue a decision in respect of the non-UPC designation(s), subject to a condition subsequent that the competent national court does not later hold the relevant patent wholly or partially invalid in a way which would affect that infringement finding.

If a competent national court does later hold the relevant patent wholly or partially invalid in such a way, the patentee will have two months from the date of that decision to request consequential orders from the UPC, including to stay the proceedings until a final decision is given by the national court.

Outcome of the case

Applying this to the facts before it, the UPC Court of Appeal confirmed that the UPC had jurisdiction over the UK aspects of the infringement claim, based on the defendants’ domicile (Germany).

It found the German designation of the patent valid and that the Sonora Xtra-3 printing plates fell within the scope of the claims. However, Kodak was ultimately able to rely on a private prior use defence to avoid infringement in Germany.

As for the UK designation, applying English law, the UPC Court of Appeal found no infringement by any of the defendants on the facts.

Comment

This is another important jurisdictional decision that helpfully builds on earlier UPC case law, and the CJEU’s decision in BSH v Electrolux, where jurisdiction is established based on the defendant’s domicile. In providing a structured framework for the exercise of the UPC’s long-arm jurisdiction in those types of cases, it should make the UPC’s role in cross-border patent enforcement more predictable for patentees and defendants alike.

That said, this decision should be seen as one piece of a rapidly-developing jurisdictional landscape, in which the precise limits of the UPC’s jurisdiction are still being determined. The UPC Court of Appeal has taken a narrower view on its jurisdictional reach where jurisdiction is founded on the place where the harmful event occurred (see Keeex v Adobe) and a number of questions remain pending before the CJEU in relation to claims pursued against non-EU defendants for acts committed in non-UPC EU territories (e.g. Spain) based on the presence of an anchor defendant or intermediary that is domiciled in a UPC contracting member state (Dyson v Dreame).

Perhaps unsurprisingly, the key area of disagreement between the parties related to price. Both parties agreed that Samsung would be the net payer under any new cross-licence, but they disagreed about the amount of that net payment. Samsung’s position was that it should not have to pay more than $200m, whereas ZTE was seeking $731m.

In order to reach a conclusion on what was FRAND, the court had to determine a number of important factors, including which comparable licence to use in the valuation process, the mechanics of unpacking that licence and repacking it into the court-determined licence, and any adjustments needed to account for non-FRAND factors (see our blog for more details).

Applying its findings to the facts before it, the court ultimately concluded that the FRAND lump sum balancing payment due from Samsung to ZTE under a court-determined licence should be $392m.

The judgment did not, however, require ZTE to enter into that licence. ZTE did not give an undertaking to the English court to take a licence on these terms and had brought its own claim in China for a determination of global FRAND terms.

The Chinese court handed down its FRAND determination at a similar time, finding, in contrast to the English court, that ZTE’s offer of $731m for a six year licence was FRAND.

A third decision on FRAND terms between Samsung and ZTE was handed down by the Munich I Regional Court in Germany, which suggested a FRAND lump sum payment of $640m.

And, finally(!), the UPC proposed somewhat of a halfway house, suggesting that the parties enter into a global settlement on the basis of either:

• Payment of a lump sum from Samsung to ZTE of $640m for a cross licence lasting until 31 December 2028; or
• Payment of a lump sum from Samsung to ZTE of $731m for a cross licence lasting until 31 December 2029,

with a suggestion that the parties submit to mediation before the UPC’s new Patent Mediation and Arbitration Centre if they are unable to agree.

That apparent push to alternative dispute resolution (ADR) seems to be a recurring theme at the moment, with the English Court of Appeal having recently held, in Acer v Nokia, that an offer of an interim licence whose terms were subject to adjustment by the decision of an arbitral tribunal was RAND. That led the Court of Appeal to permanently stay the English proceedings and left Acer and Asus with a choice – to accept the adjustable licence offer and thereby consent to arbitration; or refuse the licence offer and take the consequences (they would be seen by the English court as an unwilling licensee and therefore unable to invoke the court’s jurisdiction to determine (F)RAND terms).

Shortly afterwards (albeit following a subsequent English court decision finding one proposed term of the adjustable licence offer non-RAND), Nokia signed separate patent licence agreements with both Acer and Asus, agreeing in the process to pause or withdraw all related patent litigation and enter into binding arbitration to determine certain terms of the respective agreements.

Whilst this type of support for arbitration isn’t new, particularly in the English courts (with Lord Justice Arnold, for instance, long having been in favour of it for these types of case), these recent developments suggest we might start to see more parties taking the ADR route in future.