Transparency reporting and the streaming problem

Cloudflare has published its transparency report for the first half of 2025, continuing a twice-yearly practice it has maintained for more than a decade. The reports document how the company handles legal requests for customer information and abuse reports, and the format itself has been revised over time as the company has grown. Beyond the numbers, the first half of 2025 brought notable developments in one area in particular: unauthorized streaming and copyright claims.

Cheaper access to AI tools and other abuse systems has made bad actors more sophisticated in how they attempt to stream copyrighted content, frequently adding steps to conceal their activity. Cloudflare's answer has been to experiment with new methods for handling streaming and infringement allegations, working closely with rightsholders to identify domains and accounts that may be streaming, to respond faster in real time, and to spot possible risks earlier.

Policymakers, rightsholders and online service providers share an interest in preventing pirated streaming of sporting and other events. The actors behind unauthorized streaming may also misuse Cloudflare's services in ways that affect performance, cost and reliability for everyone else. That common ground has produced partnerships and improved information sharing — necessary, Cloudflare argues, because streamers keep finding new ways to evade detection.

An abuse process built to scale

With roughly 20% of the web behind its network, Cloudflare has long treated smart, scalable abuse handling as a requirement rather than an option. Even as a smaller company with a narrower service portfolio, it built a system that routed abuse reports to the party best placed to act on them — usually the website owner or hosting provider — without compromising that party's security.

New services have each received a service-specific approach to abuse shaped by the nature of the service, legal requirements and human rights considerations. Hosted content, for instance, is treated differently from content on sites using Cloudflare's security and CDN services, and that distinction runs throughout the report. Cloudflare also builds systems that learn from the reports it receives, since identifying abuse patterns improves both mitigation on its network and protection for customers against a broad range of cyber threats.

AI cuts both ways: bad actors use it to spin up phishing campaigns quickly and to shift and split unauthorized streaming traffic to dodge detection, and LLMs also make it easy to generate floods of low-quality or malicious abuse reports. On the defensive side, machine learning applied to the traffic and data behind Cloudflare's network has produced automated systems that can operate at the scale of the problem while more accurately separating genuine abuse from noise. In 2024, as documented in its abuse transparency report, the company expanded automated responses to technical abuse reports such as phishing, and has taken similar internal steps to identify abusive patterns before services are misused.

Streaming enforcement in the first half of 2025

Cloudflare's free plan has protected websites since 2010, on the reasoning that sites shouldn't pay for being attacked, and it works because ordinary websites consume little bandwidth. Streaming does not fit that model: a single second of typical video uses as much bandwidth as loading an entire webpage. To keep the free tier viable, Cloudflare has always restricted using free services to deliver streaming video. Most customers respect the limits, but some attempt to misconfigure the service to stream video anyway.

During the first half of 2025, Cloudflare worked with several large rightsholders on unauthorized streaming. The collaboration included an API for streamlined reporting, feedback on report quality so rightsholders supply actionable information, and — after verifying reports against internal metrics — scaled-up responses to streaming reports. The results:

  • DMCA reports for websites on Cloudflare's hosted services rose from approximately 11,000 in the second half of 2024 to approximately 125,000 in the first half of 2025.
  • Cloudflare took action on 54,000 reports, up from 1,000 in the second half of 2024, while speeding up its notice-and-takedown process.
  • Signals in those reports led to hosting services being terminated for another 21,000 accounts.

Rightsholder data also strengthened the technical tools used against unauthorized streaming by sites on Cloudflare's non-hosted services. To keep free and low-cost services available to static websites, Cloudflare may act on sites using those services when they appear to be streaming, regardless of whether the content infringes copyright. Rightsholders' reports focus on infringement, but Cloudflare uses them as signals to inform those tools. The closer cooperation has shortened response times on specific reports and helped block thousands of similar sites attempting unauthorized streaming before they were ever identified as infringing.

What Cloudflare learns about streamer tactics also feeds its wider cybersecurity work. Earlier in 2025, the company drew on information from its streaming program to help a smaller customer whose services were being abused to host streaming content without the customer's knowledge, providing guidance and tools to stop the behavior after understanding how illegal streamers were gaining access.

Cloudflare expects streamers to adapt to these measures and says its work on the problem is not finished.

Blocking orders and their limits

Where Cloudflare has pursued technical fixes in real time with rightsholders, many regulators and rightsholders have chosen legally mandated blocking of the Internet. A lack of technical expertise, or simple indifference, can cause significant overblocking of innocent websites, often with no transparency or accountability. Cloudflare sides with civil society groups such as the Internet Society that removing illegal content at the source remains the best approach.

LaLiga, the Spanish football league, is cited as a prominent example: working through ISPs in Spain, it has blocked IP addresses shared by many thousands of websites during matches, with no government oversight, causing severe Internet outages in Spain during match times. The disproportionate effect of IP address blocking is well documented. LaLiga has been unapologetic about blocking countless unrelated websites, Cloudflare says, effectively asserting that its commercial interests outrank Spanish users' right to access the broader Internet during matches. The Spanish government has not acted to protect those users, and Cloudflare has seen no concrete evidence that such blunt-force blocking meaningfully solves the problem.

In Cloudflare's view, regulators and rightsholders should seek proportionate ways to prevent infringement and work collaboratively with service providers rather than damage the Internet's foundations. Infrastructure-layer blocking tends to be overbroad, non-transparent and ineffective. Still, the company has examined more targeted forms of blocking, finding in general that it is of limited effectiveness because determined users find workarounds. Cloudflare has complied with valid orders concerning its CDN services that satisfy human rights principles on proportionality, due process, free expression and transparency; in countries whose laws provide for blocking with appropriate oversight, it may geoblock websites to limit access within the relevant jurisdiction through its CDN services.

Cloudflare has never blocked through its public DNS resolver, arguing that such demands conflict with an open Internet and would require tools contrary to the resolver's design; it continues to litigate against efforts to force it to build those capabilities. It has sometimes geoblocked access through its CDN and security services in response to DNS blocking orders.

The first half of 2025 saw a marked increase in blocking orders in Europe. Private rightsholders obtained multiple orders directing Cloudflare to block websites in Belgium, France and Italy; Cloudflare challenged aspects of them while geoblocking the sites at issue in the relevant countries through its CDN and security services. Cloudflare also began giving effect to UK court orders directing other service providers to block websites identified as dedicated to copyright infringement. Under a voluntary agreement with rightsholders, it geoblocks those sites through its pass-through CDN and security services. Visitors to affected domains see an interstitial page returning a 451 status code that points to the specific order and includes a process for affected parties to contest the action.

BLOG-3078 image 1

The UK approach reflects a desire to experiment with a more targeted method than the overblocking seen elsewhere in Europe, and rests on the UK regime's protections around proportionality, due process and transparency, including redress for affected parties. Cloudflare is monitoring the impact and has reserved the option to change course if the system is abused.

Blocking demands also reached new subject areas. Official government notices from France and Belgium reported that websites on Cloudflare's hosted services were offering gambling services illegally in those jurisdictions. In both cases, Cloudflare passed the notice to the customer, which took its own action to resolve the issue — an illustration of the value in connecting customers directly with regulators instead of proceeding straight to a blocking demand.

What comes next

Cloudflare says it will keep working with rightsholders and regulators toward effective, proportionate responses to online abuse, and will continue to use its biannual transparency reports to set out the principles behind that work and behind its handling of abuse reports and customer information requests.