Cases are increasing in Europe where DNS servers or VPNs that people were using to watch sports broadcasts or overseas dramas suddenly become inaccessible. "Collateral blocking"—where communication blocking intended to combat piracy ensnares unrelated users—is becoming routine there. Amid this, on June 25, 2026, Google submitted a 21-page comment to the public consultation on the review of the EU Copyright Directive (DSM Directive), opposing the blocking of DNS resolvers, IP addresses, and VPNs.
But this submission came three months after Google itself had its challenge to a blocking order rejected by a French court. The figures on collateral damage that Google lined up as grounds for opposition are not unrelated to Google's own litigation.
June 25: The 21 Pages Google Presented to the European Commission
On June 25, 2026, Google submitted a 21-page, 456-kilobyte comment—recorded as Feedback ID F33504634—to the European Commission's "Have Your Say" public consultation (review of the 2019 DSM Copyright Directive, Initiative ID 18172). The submitter was listed as Google (in the large enterprise category, 250+ employees), the contact person as Annelise Badinand, and the EU Transparency Register number as 03181945560-59. The consultation opened on May 18, 2026, with a deadline of June 25. Google's submission arrived on that very deadline day.
The scope of this consultation is broad. It covers issues ranging from copyright licensing challenges posed by generative AI, online piracy countermeasures, remuneration for performers and producers, to access to copyrighted works for research purposes. Google's submission focused on one corner of the piracy countermeasures issue: the blocking of DNS resolvers, IP addresses, and VPNs. In its comment, Google reportedly argues that these measures are ineffective because they "remove no content whatsoever and are easily circumvented by using a different DNS resolver." It also states, according to TorrentFreak's reporting, that such measures are "disproportionate, ensnare lawful services, raise extraterritoriality concerns, and block entire domains."
The real-world examples Google cites span multiple countries including Portugal, Italy, France, and Spain. The oldest example dates to December 2019 in Portugal. When a local ISP blocked one of Google's hosted virtual IP addresses, Google recounts in its comment that it "disrupted core Google-related services and blocked legitimate traffic from other legitimate Google Cloud customers sharing the same virtual IP." Among these, the three cases from Italy, France, and Spain come with actual measured data showing the scale of collateral impact. The reason blocking aimed at shared infrastructure ensnares unrelated users lies in the difference between two mechanisms: DNS resolvers and IP addresses.
Why Does One Block Ensnare Millions of Domains?
A DNS resolver is a system that converts a domain name entered by a user into a server's IP address. Major DNS resolvers such as Google Public DNS and Cisco OpenDNS handle a massive volume of queries for countless domains worldwide. When authorities want to block a pirate site, a common technique is to stop queries for that domain at the DNS resolver level. But when subdomain-level misjudgments or system malfunctions occur, unrelated services get caught up as well.
Indeed, in October 2024, Italy's anti-piracy system "Piracy Shield" mistakenly designated a Google Drive subdomain for blocking. The block reportedly lasted about six hours. Piracy Shield is a system that automatically adds domains and IP addresses to a blocklist based on reports from rights holders with broadcast rights; its defining feature is immediacy without human pre-screening, but that speed comes with an inherent risk of misjudgment.
IP address blocking tends to produce even larger-scale collateral damage. CDN (content delivery network) operators such as Cloudflare and Akamai adopt a "shared IP" approach, assigning millions of domains to a single IP address for cost efficiency. On February 24, 2024, Piracy Shield designated Cloudflare's shared IP address "188.114.97.7" for blocking. According to IPinfo statistics at the time, 42,243,794 domains were associated with the entire Cloudflare network to which this IP belonged (Autonomous System Number AS13335).
This figure does not represent the number of domains directly hanging off the single blocked IP. Rather, it is a yardstick showing the potential scale of collateral damage that Cloudflare's shared-IP approach can produce. Even so, multiple reports have confirmed that sites belonging to unrelated organizations—such as the charity ODV Prison Volunteers Association, which supports communication between inmates and their families, and the telecom company Elimobile.it—actually became inaccessible.
There is a reason authorities choose IP-address-level blocking. Identifying individual pirate pages and blocking them at the domain level requires rights holders to continuously monitor infringing content and update blocklists as needed—a labor-intensive process. Blocking a shared IP address en masse skips that effort and produces immediate results. But for the same reason, unrelated domains that happen to share that IP address get blocked indiscriminately. Speed and precision exist in a trade-off relationship, and the risk of collateral damage grows larger the more a system, like Piracy Shield, leans toward instant blocking.
In France, following a complaint by broadcaster Canal+, blocking orders were issued against the DNS resolvers of Google, Cloudflare, and Cisco. In response, Cisco discontinued providing its OpenDNS service in France entirely, effective June 28, 2024. Rather than merely blocking individual domains, this was a more drastic step of halting the service offering itself.
In Spain, collateral blocking via IP blocking has recurred every time the football league LaLiga airs live broadcasts. According to a survey by the research organization Open Observatory of Network Interference (OONI), which examined 9.2 million domains from within Spain between January and June 2026, at least 554,507 domains (roughly 5.8% of those surveyed) were blocked at least once. The affected IP addresses numbered 7,441, spanning 36 organizations including Amazon, Cloudflare, Alibaba Cloud, Akamai, Meta, and Microsoft. This means the IP blocking carried out every time LaLiga broadcasts have repeatedly ensnared the cloud services of these unrelated companies.
The Advocate Opposing Blocking Was, in Fact, a Party to a Lost Lawsuit
On March 27, 2026, the Paris Court of Appeal dismissed all five challenges brought jointly by Google, Cloudflare, and Cisco. At issue was the legitimacy of a DNS blocking order based on Canal+'s complaint, and the court ruled that DNS blocking was "technically feasible and proportionate." The court also rejected Cisco's argument about the high implementation cost of geo-restricted DNS filtering, finding it "lacking objective grounds."
Google submitted its comment to the European Commission on June 25—three months after this dismissal. Google is thus both a party submitting comments to the European Commission questioning the legitimacy of DNS blocking, and a party that, on this very same issue, recently lost in a French court. Neither the comment itself nor the media coverage to date explicitly emphasizes this conflict of interest.
The fact that the motive may be self-protective does not immediately negate the technical validity of the points raised. The scale of collateral risk inherent in Italy's shared-IP blocking, and the figure of 554,507 mistakenly blocked domains in Spain, can both be confirmed through OONI's independent survey and IPinfo's data. Google's comment does not contradict these independent measurements, and even when an interested party makes a claim, the reliability of the underlying data itself can be assessed by separate standards.
Simultaneous Regulatory Expansion and Judicial Pushback on Both Sides of the Atlantic
On the other side of the Atlantic, a tug-of-war over blocking is unfolding in parallel. The $1 billion damages award ordered against Cox in a 2019 jury verdict had already been vacated by the Fourth Circuit Court of Appeals in 2024, which remanded the case for recalculation of damages. What the U.S. Supreme Court overturned on March 25, 2026, was the very finding of contributory copyright liability that had served as the foundation for that award. While all nine justices agreed on the conclusion in favor of Cox, the majority opinion (joined by seven justices) that drove the ruling established a standard limiting ISP contributory liability to cases where the ISP "actively induced infringement" or "tailored its service to facilitate infringement" (Justices Sotomayor and Jackson concurred only in the judgment, not in this standard). This ruling narrows the judicial path toward expanded blocking.
When the judicial route narrows, the legislative route moves next. In mid-2025, TorrentFreak reported on the contents of a discussion draft, the "American Copyright Protection Act (ACPA)," led by Representative Issa (R-CA) in the U.S. House. The draft would require ISPs and DNS resolvers to comply with blocking orders using "all reasonable measures," while exempting DNS providers with fewer than 50,000 annual users and ISPs with 1% or less of the U.S. market share.
This bill is called an "alternative site-blocking route" not in the sense of circumventing DNS blocking, but because it represents new domestic U.S. blocking legislation—the first since SOPA's demise in 2012. If anything, it aligns with the EU discussion in including DNS resolvers as targets of orders. On June 30, 2026, the House Judiciary Subcommittee on Intellectual Property held a hearing on copyright protection and internet enforcement, and Chairman Issa indicated his intention to introduce a site-blocking bill within the week, stating that "the final compromise language has been worked out." Whether this bill directly carries forward the 2025 ACPA discussion draft could not be confirmed at the time of writing, but it is clear that preparations for similar legislation in the House have entered their final stage. Google's submission came just a few days before this hearing.
A similar dynamic occurred once before, in 2012. At that time in the U.S., a bill called SOPA (Stop Online Piracy Act), which included DNS blocking provisions, was under deliberation, but was scrapped following blackout protests led by Google and Wikipedia. Fourteen years later, the conflict between tech companies and rights holder groups has flared up again over the same technical issue: DNS blocking. What differs is that, while the bill itself was buried back then, this time it is the Supreme Court that has narrowed the judicial route, prompting rights holders to shift their focus toward the legislative route.
Sorting out the interests at stake: those who stand to benefit from regulatory rollback are the Google/Cloudflare/Cisco camp and EuroISPA, the industry group representing more than 3,300 ISPs across Europe. In June 2026, EuroISPA submitted a comment to the European Commission calling for rights holders to bear financial responsibility for collateral damage caused by excessive blocking, citing an April 2026 CEPS (Centre for European Policy Studies) study as its basis. Conversely, those who stand to lose from regulatory rollback are the broadcast rights holder camp, including LaLiga and Canal+, and Italy's AGCOM (the communications regulator that operates the Piracy Shield system). If blocking measures are restricted, the very effectiveness of anti-piracy enforcement could be called into question.
What Japan Set Aside in 2018 Is Now a Reality in Europe
Japan once faced this same issue. In 2018, the government's Intellectual Property Strategy Headquarters considered legislating site blocking, but opinions for and against were divided, and it was concluded that "no consensus could be reached in this study group." The focal point was the secrecy of communications guaranteed under Article 21 of the Constitution—the concern that blocking, ostensibly targeting specific pirate sites, was technically inseparable from a mechanism that censors the entirety of communications content.
The secrecy of communications refers to the right, established under Article 21, Paragraph 2 of the Constitution, to prevent the content of communications and information about the parties involved from being known to the state or third parties. To implement DNS blocking, the resolver must grasp which domain a user is attempting to connect to and cross-reference it against a blocklist. The argument that this cross-referencing process itself could violate the secrecy of communications became the central basis for opposition in the 2018 study group.
This conclusion still stands today. Even in the government's "Comprehensive Countermeasures Menu Against Online Piracy," revised in May 2024, the establishment of a legal framework for blocking is described as something to be considered while observing the effectiveness of other measures—meaning legislation continues to be deferred.
What is happening in Europe now can be described as a case where the concerns Japan debated in theory back in 2018 have materialized in concrete numerical form. Blocking a single shared IP address carries a collateral risk comparable in scale to Cloudflare's entire network (42.24 million domains), and over 550,000 domains actually get mistakenly blocked every time a football broadcast airs. Japan's debate, which set aside legislation citing the secrecy of communications, took place without such measured data on collateral damage. Should Japan revisit this issue in the future, Europe's real-world examples would provide material that could be cited by either the pro- or anti-blocking side.
The Turning Point: EuroISPA's Financial Liability Proposal and the Conditions Set by Independent Data
Alongside this review, the European Commission is also soliciting comments on the possibility of targeted legislative measures, and it remains unclear whether it will move toward revising the Directive or limit itself to more narrow measures—neither the direction nor the timing has been indicated yet. Among the factors that will influence this decision, the hardest to gauge is where EuroISPA's proposed financial liability would fall. This industry group, representing more than 3,300 ISPs across Europe, proposes—citing an April 2026 CEPS study as grounds—a mechanism that would place the cost of collateral damage on rights holders. While Google's comment carries the weakness of self-interest, EuroISPA's proposal represents a collective claim from an industry body, carrying weight that the European Commission cannot easily dismiss.
How far the U.S. side develops the ACPA discussion draft into concrete form could also affect the European Commission's decision. If the draft enters congressional deliberation while retaining a design that includes DNS resolvers as targets of orders, there remains a possibility that the regulatory frameworks on both sides of the Atlantic will diverge in opposite directions—just as the European Commission moves to exclude DNS and VPN blocking from the scope of the Directive.
The independent data accumulated by OONI and IPinfo already exists, separate from Google's self-interest as a party. Given this accumulation of independent data, together with the narrowing of the judicial route in the Cox v. Sony ruling, there is a somewhat higher likelihood that the European Commission will lean toward excluding DNS and VPN blocking from the scope of the Directive and placing the burden of proof on rights holders. However, this shift would be conditioned on whether rights holders can present a low-misfire alternative to Piracy Shield's blocking method. If they cannot, there remains a possibility that instant-blocking systems will instead be preserved as the standard.