France's media and digital regulator has crossed a meaningful threshold in how democracies police the live internet. Beginning May 18, 2026, ARCOM trialled real-time blocking of pirate IPTV servers' IP addresses during the Roland-Garros tennis open — the first such test at a major French sporting event. During matches, rights holders detect pirate streams and report them to ARCOM, which orders France's four main ISPs — Orange, SFR, Bouygues Telecom, and Free — to block the offending IPs immediately, with the order lifted when the match ends.
This is a real shift in technique, not just scale. Until now, France's anti-piracy enforcement under Article L. 333-10 of the Code du sport — the "dynamic injunction" regime created by the October 2021 anti-piracy law — worked mostly at the domain-name layer: courts order blocking, ARCOM's sworn agents identify new mirror sites, and ISPs null-route the domains. The move from blocking domain names to blocking server IP addresses in real time is what's new at Roland-Garros, alongside a published recommendation, two model agreements, and a centralized blocklist system ARCOM unveiled on May 22, 2026.
The case for moving fast
The strongest argument for ARCOM's approach is straightforward, and it deserves to be stated plainly. Live sport is the one product whose value collapses the instant it is freely available — a blocked pirate stream at minute 70 of a final is worth far more to a broadcaster than a court order issued three days later. DNS-layer blocking has been, in the regulator's own framing, too slow and administratively heavy to matter for live events; pirates simply rotate domains faster than the paperwork can follow. ARCOM reports that illegal sports broadcasts caused roughly €290 million in direct damage in 2024, and that 6,496 services were subject to blocking or delisting requests in 2025. For a sector that funds the rights fees underpinning much of professional sport, that is not a trivial leakage, and "do nothing faster" is not a serious policy.
ARCOM also built in genuine safeguards, which distinguishes this from a blunt kill-switch. There is an untouchable-IP whitelist (notably covering ISP infrastructure), an initial prohibition on blocking shared servers, blocking windows limited to the duration of the broadcast, and legal responsibility resting on the rights holder that requested the block if it goes wrong. These are not cosmetic; they are precisely the controls that an earlier generation of blocking orders lacked.
Why proportionality is the whole ballgame
And yet the design still runs against a hard technical fact: an IP address is not a website. A single address routinely hosts a pirate stream alongside thousands of unrelated, perfectly legal services, because modern hosting, CDNs, and reverse proxies multiplex many tenants behind one address. Block the IP and you block everything behind it.
This is not a hypothetical. Spain offers a fully-run experiment. Throughout 2024–2025, IP-blocking orders obtained by LaLiga and executed by Spanish ISPs repeatedly took down infrastructure belonging to Cloudflare, and at various points rendered GitHub, Docker, and Vercel unreachable for Spanish users — with well over 3,000 legitimate sites reported caught in the net. The Spanish courts largely waved this through as acceptable collateral damage. The lesson is not that real-time blocking is impossible to do carefully; it is that the temptation to trade precision for speed is structural, and that once the machinery exists, the blast radius tends to grow.
France's whitelist-and-no-shared-servers posture is a direct response to the Spanish debacle, and it is welcome. But it also quietly concedes the core problem: the most dangerous targets — pirate operations hiding behind the same CDNs that serve half the legitimate web — are exactly the ones the safeguards carve out. An enforcement tool that works cleanly only against pirates foolish enough to use dedicated, easily-isolated servers will push the rest deeper into shared infrastructure, where blocking them means breaking everyone else.
The accountability gap
There is also a governance concern that the DSA era makes sharper. Under L. 333-10, a judge sets the framework, but the moment-to-moment decisions — which IP, blocked when, for how long — are delegated to a rights-holder-to-ARCOM-to-ISP pipeline operating on sub-hour timescales. Civil-liberties groups including La Quadrature du Net have long warned that this administrative fast-track erodes the per-block judicial scrutiny that blocking a communications channel should require. Real-time automation, with broad rollout targeted within six months and reports of plans to handle up to 10,000 IPs at once for the World Cup, widens that gap between the speed of blocking and the speed of redress. A wrongly-blocked legitimate business has no realtime appeal; by the time anyone notices, the match — and the harm — is over.
A proportionate path
None of this argues for tolerating live piracy. It argues for keeping the tool inside the proportionality envelope that L. 333-10 itself demands. That means: hard, audited adherence to the no-shared-server rule rather than case-by-case erosion; mandatory public transparency reporting on every IP blocked and every collateral outage; a fast, real-person redress channel for wrongly-blocked services; and a sunset clause forcing legislative review before any permanent, automated, ten-thousand-IP regime goes live. France has built safeguards Spain did not. Whether they hold under the pressure of a World Cup final — when the incentive to block first and ask later peaks — is the test that actually matters.