DPG Media et al vs. HowardsHome – the second instance Dutch ruling on the press publishers' right and TDM exceptions

Introduction

The second instance ruling in DPG Meda v. HowardsHome has been delivered on 20 October 2026 by the Amsterdam Court of Appeal (“Court of Appeal”).

The first instance ruling in late 2024, previously discussed here, was the first in the EU on the “very short extracts” exception of the press publishers’ right (“PPR”) introduced with the 2019 Copyright in the Digital Single Market Directive (“CDSMD”). This exception was not in scope of the preliminary questions about PPRs answered in the CJEU judgment in Meta (Case C-797/23, discussed here and here), nor will it likely surface in the highly anticipated CJEU ruling in Streamz (Case C-663/24). So, to the best of our knowledge, this Dutch appeal ruling is for now the highest court ruling to shed light on it.

In addition, the initial Dutch ruling was one of the first to elaborate on the appropriate manner of exercising opt-outs in compliance with the text and data mining (“TDM”) exception in Article 4 CDSMD. However, the Court of Appeal now ruled that the activities at hand do not constitute TDM. Therefore, it did not elaborate on the appropriate manner of opting out, nor did it see grounds to refer preliminary questions to the CJEU on this matter.

Facts

The plaintiffs are the Belgian media companies DPG Media and Mediahuis (together: “the Publishers”). The defendant is the company HowardsHome, which ceased all its operations in 2024.

The Publishers offer national and regional dailies that also include websites, both dominant in the Dutch online commercial news sector with a combined market share of over 90%. The defendant, which stopped providing its service in the course of the dispute, offered a paid alert service under the name “HowardsHome Nieuws” (“HowardsHome”) to public institutions and private companies since 2001. In its most recent activity, HowardsHome obtained the information for its alerts mainly from Really Simple Syndication-feeds (“RSS-feeds") of news items publicly available online. RSS-feeds contain various information about an article. In its alerts to its customers, HowardsHome provided a hyperlink to the relevant post, including the title, a short description and, if publicly available, a thumbnail. HowardsHome placed these alerts on a digital portal.

The alerts included news items from the Publishers. Among other things, HowardsHome allegedly offered to its customers content of the Publishers that was not obtained through purchasing, but from RSS-feeds available on the internet or obtained by scraping websites. The Publishers argue that this infringes their exclusive rights to reproduction and to making available of their work to the public.

At its core, this dispute is about whether HowardsHome's alert service infringes the Publishers' PPR, copyright and database rights. In first instance, HowardsHome successfully relied on the PPR exception for “very short extracts”, as well as on the Article 4 CDSMD TDM exception for copyright infringement. The Court of Appeal overturned both those findings, while also ruling that HowardsHome cannot successfully invoke any other statutory limitation on copyright (paras. 6.14–6.20).

Press publishers’ right

The first instance court ruled that HowardsHome’s snippets of 150 characters in its RSS-feed alerts, amounting to approximately 20 words, fall within the “very short extracts” exception of the PPR. On this point, the Court of Appeal first stated that “it is not possible to provide a general criterion for answering this question, such as a maximum number of characters or words that applies in all cases” (para. 6.39). However, in the light of the circumstances of the case, it found that the excerpts from HowardsHome, even if they were limited to a maximum of 150 characters (excluding the headline), were too long to be regarded as “very short extracts” (para. 6.42). It also noted that the excerpts caused a “sufficiently concrete threat” to the Publishers’ revenue model (para. 6.24). Finally, it considered that HowardsHome cannot justify the length of its snippets by invoking the right to freedom of information, since reports on such a scale are not necessary for that purpose (para. 6.43).

Perhaps unsurprisingly, the PPR here functions as another stick for rightholders to drive smaller competitors out of the market, whereas it remains unclear in practice to what extent this right actually serves its intended purpose. That is, increasing the bargaining power of publishers vis-à-vis the Googles and the Facebooks of the world, who have long benefited from publishers’ news articles without properly remunerating them.

In principle, it does not seem unfair to require the commercial service of HowardsHome to operate under a license either. Yet a more extensive listing of the facts suggests the company was willing to enter into a new licensing deal with the Publishers. Contrary to the first instance court’s position, the Court of Appeal did not consider that the Publishers made a license impossible through imposing unrealistic conditions on HowardsHome and charging exorbitant prices. Unfortunately, the competition law aspects that seem highly relevant in this context were not addressed in the ruling either, due to a failure to meet the burden of proof by HowardsHome.

The TDM exception in Article 4 CDSMD

Also contrary to the first instance decision, the Court of Appeal ruled that HowardsHome’s activities cannot be regarded as TDM. The Court of Appeal recalls those activities as follows. Through its alert service, HowardsHome focused on selecting news articles suitable for its clients. In this way, using the text and, where applicable, the photographic material from the original press publications, it generated alerts for its clients. According to the Court of Appeal, such activities do not fall under the scope of TDM defined in Article 2(2) CDSMD as “any automated analytical technique aimed at analysing text and data in digital form in order to generate information which includes but is not limited to patterns, trends and correlations” (para. 6.24).

In public commentary on the ruling in first instance, it did not raise much concern that the court accepted HowardsHome’s activities to fall within the definition of TDM without further elaboration. On closer inspection, admittedly, it is debatable whether the court was right to do so, even with as broad a wording as the TDM definition provides. Surely, the first element in the definition of “any automated analytical technique” [emphasis added] creates leeway for the crawling and scraping of RSS feeds to be regarded as “analysis”. However, it is less clear that the short news overviews provided by HowardsHome, or information generated earlier in the process, constitute information which includes “patterns, trends and correlations” resulting from a form of an automated analytical technique. The Court of Appeal did not elaborate on the flexibility that the formulation “not limited to” in relation to the three examples could allow for. HowardsHome’s portal allowed customers to filter and sort the collected news articles by topic, date and keywords, amongst other things. Assumedly, the Court of Appeal did not consider such information sufficient to fall within the scope of the TDM definition.

Conclusion

This Dutch second instance ruling shows how narrowly the PPR’s “very short extracts” exception can be interpreted by courts. The length of the excerpts of news articles shown in this case does not differ from what a search engine provides. This points to the unfortunate situation exacerbated by the introduction of PPRs that large platforms increasingly are the only ones able to afford the licence fees charged by leading publishers, whilst smaller services, like HowardsHome, fade into the background.

Furthermore, the Court of Appeal did not consider HowardsHome’s process of providing automated news alerts through scraping RSS-feeds as TDM activity in the sense of the CDSMD. With that, the ruling primarily offers insights limited to a process that has elements of TDM, i.e. web scraping, but has markedly different purposes than the TDM done in relation to training AI models. The latter is not the only activity relevant for the TDM exception in the CDSMD (see here), but will remain central in current copyright debates, perhaps even more so after the ruling discussed in this contribution. 

 Image by Hilda S via Pexels