Paris court on site blocking: reasonable, not effective

The Paris Judicial Court refused daily penalties against a CDN provider on 17 September, holding that blocking must be reasonable rather than effective.

On 17 September 2026 a panel of three judges at the Paris Judicial Court refused a French pay TV broadcaster's request for financial penalties against a CDN and DNS provider over pirate sport streaming sites, according to a report. The broadcaster had asked in May for €50,000 per day for every site that remained accessible, the same again for each new site reported to the provider, and €20,000 in legal costs. All of it was refused.

The reason is the part worth reading. The court read an earlier EU Court of Justice ruling on blocking injunctions as requiring intermediaries to take reasonable measures, not to guarantee that blocking works.

If you hold live sport rights, that sentence is the most consequential thing in the story, and it has nothing to do with the provider.

What happened

This was decided at first instance and the broadcaster can still appeal, so nothing here is settled law.

Two points of fairness that the coverage of rulings like this usually loses. The provider was ordered to block through its CDN, it did so, and the court treated that as evidence of genuine cooperation. It was not penalised and it is not a wrongdoer in this story. The provider told the court its resolver's architecture does not allow blocking, and its own transparency report states it has not blocked content through that resolver.

Separately, a similar and potentially much larger penalty request by another sports broadcaster against a non profit DNS resolver is still pending, with a Paris decision that the report, published on 8 October 2026, expected later in October. That matter is not decided and nothing here should be read as if it were.

What the news actually is

The headline most people will take from this is "CDN provider wins in France". That is true and it is not the story.

The story is the standard. A court has stated, in terms, that an intermediary's obligation is to take reasonable measures rather than to achieve an effective block. The practical meaning of that is a question most rights holders have never had answered so directly: if the order does not work, whose problem is that?

The ruling answers it. It is the rights holder's problem. The court held that the provider is not responsible when sites switch to a different CDN or redirect to mirrors. The broadcaster must approach the new intermediary itself, and report new mirror sites to Arcom, whose role in this is to receive reports of infringing sites.

So the enforcement does not end with the order. The order is the point at which the ongoing work transfers to you.

The mechanism: the file never moves, only the route does

The reason that transfer is so expensive comes down to what blocking actually touches.

Only the last row changes what is stored. Everything above it changes a route. A stream being served from an origin server does not care which path the audience takes to reach it, and the people operating it can change that path in minutes by moving CDN or standing up a mirror. The rights holder then has to detect the change, identify the new intermediary, and start again.

That asymmetry is the whole economics of it. Changing a route costs the operator a few minutes. Responding to a changed route costs the rights holder a detection cycle, a legal contact and, in the French arrangement, a fresh report to a regulator. The same pattern applies to visibility work, and our guide to filing a copyright complaint with a search engine sets out what delisting does and does not achieve for exactly this reason.

The gap the reader cannot see

Here is the part that never reaches a budget line.

An enforcement programme built on blocking orders is usually costed as legal work: counsel, filings, the occasional hearing. Those costs are visible, they are forecastable, and they end when the order is granted.

The cost this ruling confirms is an operations cost, and it is permanent. Somebody has to watch for CDN changes, watch for mirrors, maintain contact routes into a growing list of intermediaries, and file reports continuously, for as long as the rights are worth protecting. It is not a project. It has no completion date. And it is almost never staffed, because the win that created it was recorded as a legal victory rather than as the start of a standing obligation.

For live sport the gap is sharper still, because the asset expires. A stream that is blocked twenty minutes after kick off was not protected in any meaningful sense. A standard that asks only for reasonable measures sits comfortably alongside a match that was fully available for its entire duration, and both statements can be true at once.

It is also worth being careful with the numbers in this ruling, because they illustrate the problem rather than resolve it.

The report notes the ruling does not explain how the first figure relates to the second and third. A modelled or aggregate effectiveness percentage and a count of blocked domains are not the same measurement, and nobody should treat the 72.6 per cent as a measured outcome of anything without knowing what went into it.

What this does and does not change

• It creates no new law. It is a first instance interpretation of an existing EU standard.

• It imposes no new obligation on any intermediary. The provider's existing CDN blocking was found sufficient.

• It makes no finding of wrongdoing against the provider, which cooperated and was not penalised.

• It does not require anyone to block through a public DNS resolver, and it did not test whether that is technically possible beyond accepting the provider's account.

• It does not decide the matter between the other sports broadcaster and the non profit resolver, which is separate and still pending.

• It does not settle the question. The broadcaster can appeal, and until any appeal is resolved this is one court's reading.

• It does not remove a single stream. No part of the decision required any host to delete anything.

• It does not award the broadcaster its costs.

Those negatives are the difference between a note your legal team can circulate and a summary that falls apart the first time somebody reads the ruling.

What a rights holder can do now

1. Cost blocking as an operation, not a case. Before the next order, write down who will watch for CDN changes and mirrors, how often, and what their weekly throughput is. If that line is empty, the order creates work nobody owns.

2. Measure the window, not the outcome. For live events, the only number that matters is minutes of availability during the event. Blocked at full time is not a result, and an aggregate effectiveness percentage will hide it.

3. Separate route changes from content changes in your reporting. Count CDN switches, new mirrors and new domains separately from removals. Collapsing them into one enforcement figure is how a programme loses sight of whether anything came down.

4. Keep a live intermediary map. Host, CDN, registrar and payment route for every target you are pursuing, kept current. The ruling makes finding the new intermediary your job, so the map is the job.

5. Budget for recurrence explicitly. The arithmetic of one order and of a stream that relocates weekly are different businesses. Our breakdown of what a DMCA takedown really costs is the version with recurrence included.

6. Pursue the host in parallel, not instead. Route level measures and host level removal are complementary. The mistake is running only the first and reporting it as protection.

How Ceartas approaches it

Our position here is specific and limited. We monitor for a rights holder's material across the sites, streams, file hosts and link aggregators where it actually appears, rather than only where it ranks or where an order happens to reach. When we find it, the request goes to the host, with the objective that the content stops existing at that location rather than becoming harder to route to. A delisting or a block is not reported as a removal, because it is not one.

The ongoing half is the part this ruling puts a spotlight on. Material that an audience wants comes back, under a new domain, a new host or a new route, so monitoring continues after the first action and the request is re issued each time. That is the permanent operation the court has just confirmed sits with the rights holder, and it is the thing the engine exists to carry.

We cannot promise that any host, CDN or registrar acts straight away, because their processing times are theirs. What we can say is that route level measures leave the file where it is, and that the work of finding the next route is now, explicitly in French law at first instance, yours. For agencies and rights teams running this across a portfolio, that is the cost worth planning around.

Stop funding orders and start funding removal

The appeal may change this reading, and the separate resolver decision may complicate it further. Neither will change the mechanics. A blocking order redirects traffic, the operator changes the route, and somebody has to notice and start again, every time, for as long as the rights are worth anything.

If you would rather that somebody was not your legal team on a Monday morning, see how Ceartas finds infringing streams and files, requests removal at the host, and re issues the request on every reappearance.

This story is part of our site blocking and copyright enforcement tracker, which follows every new law, ruling and platform move on the topic.