Permalinks, Framing and CMI: What Emmerich v. Particle Media Means for Publishers
On August 27, 2026, the Fifth Circuit answered a question no other federal appeals court had answered before: can a URL count as copyright management information under the DMCA?
The answer is not the clean yes that some publishers hoped for, and it arrives bundled with a second holding that cuts the other way. The same panel scrapped the twenty-year-old "server test" for framing and embedding — replacing it with a rule that makes it harder, not easier, to stop a site from displaying your content inside its own frame.
In short
A publisher URL is not automatically copyright management information (CMI) under the DMCA. In Emmerich Newspapers, Inc. v. Particle Media, Inc., No. 25-60550 (5th Cir. Aug. 27, 2026), the Fifth Circuit held only that URLs are not categorically excluded from CMI — qualification depends on facts the district court has not yet decided, including whether the URL clearly identifies the work or rights holder, remains stable, and is structured to communicate copyright information rather than simply to route a request.
The court reached that conclusion on an interlocutory appeal, meaning it answered two certified questions of law and sent the case back for fact-finding. Neither holding below has been applied to any specific URL yet.
First, the "server test" from Perfect 10 v. Amazon — which has shielded framing and embedding across the internet for nearly two decades — is not the right standard under the Copyright Act. The court replaced it with a "transmit requirement": a site only infringes the display right if it actually transmits the copyrighted work, not merely if it directs a user's browser to request it from someone else's server.
Second, the DMCA does not categorically exclude URLs from being copyright management information (CMI). But the court set a real bar: a URL must clearly convey the information the statute requires — title, copyright owner, or a genuine link to a page that displays that information — and not simply function as a locational address. Most permalinks, as currently structured, will not clear that bar.
Both holdings answer only questions of law. The case is remanded to the district court for fact-specific findings. Nothing is decided yet about whether Emmerich's own URLs qualify, or whether Particle removed anything unlawfully.
What actually happened between Emmerich and NewsBreak
Emmerich Newspapers publishes more than two dozen local titles across Mississippi, Louisiana and Arkansas. Particle Media operates NewsBreak, a news aggregation app and website. NewsBreak showed Emmerich's articles two different ways, and each way produced a different legal claim.
The first is "Framed View": a user taps a headline in the NewsBreak feed, and the app opens an in-app browser frame that pulls the live page directly from Emmerich's own server. Emmerich's argument was that users could not tell they had left NewsBreak without noticing the URL. Particle's counter was that its frame added only a small logo and share buttons, without hiding the publisher's own branding underneath.
The second is "Full-Text View": between 2019 and 2021, an Android bug caused NewsBreak to reproduce full article text under NewsBreak's own URL rather than Emmerich's. Particle says the glitch was fixed once flagged. A separate, earlier ruling in related litigation had already found that specific conduct unlawful on fair-use grounds — that part was not before the Fifth Circuit. What was before it: did displaying Emmerich's content under NewsBreak's own URL strip out CMI in violation of the DMCA?
The district court said no to both underlying theories — framing was not a display-right violation under the server test, and URLs categorically cannot be CMI. Emmerich got both questions certified for interlocutory appeal. The Fifth Circuit reversed the first holding's reasoning (while reaching a similar practical result) and reopened the second.
Keep the two claims separate when citing this case. The CMI dispute concerns only Full-Text View — Emmerich's argument that removing its URLs in favor of NewsBreak's own violated the DMCA. Framed View was litigated purely as a display-right question under the Copyright Act. The opinion does not hold that framing under a third party's own address is a DMCA violation; that theory was never before the court.
Framing: from the server test to the transmit requirement
The server test asked where a copy was stored. If the embedding site never held a copy on its own server, it wasn't "displaying" anything — full stop. That rule made linking, embedding and framing largely risk-free for two decades, because the Ninth Circuit's 2007 Perfect 10 decision built the reasoning around the statutory word "fixed."
The Fifth Circuit found that reasoning too narrow. It went back to the full statutory chain — display, show, copy, fixed, publicly, transmit — and concluded the operative word is "transmit," not "fixed." Under the new test, the question is who actually sends the content beyond the place it started: the site with the frame, or the site whose server responded to the request and delivered the page.
The court's own analogy: an embedding site behaves like a switchboard operator. It can connect the call, but only once the other end picks up and agrees to be patched through. If Emmerich's server is the one answering the request and delivering the page, Emmerich — not NewsBreak — is the party "transmitting" the content, even though NewsBreak's frame is what the user sees.
Read carefully, this is not simply a win for aggregators or a loss for publishers. It is a rule that rewards technical control. A publisher whose server refuses unauthorized requests — via paywall, via a blocked user-agent, via a not-permitted crawl directive — keeps its display right intact, because there was never a permitted transmission to hide behind. A publisher whose server answers every request indiscriminately has, in the court's framing, implicitly authorized what happens next. The opt-out architecture of the open internet did the legal work here, not the branding inside the frame.
For any publisher weighing whether an AI browsing agent's live-fetch-and-display behavior infringes a display right, this is the test that will now apply in Texas, Louisiana and Mississippi — and the one other circuits will be asked to consider next. An agent that only retrieves and displays what your own server was willing to transmit will be difficult to reach through the display right alone. That is precisely why the CMI question, and what travels inside the permalink itself, becomes the more durable lever.
CMI: why most permalinks will not qualify as drafted
Emmerich argued its URLs qualified as CMI three separate ways: the domain name identified the copyright owner, the article slug functioned as a title, and the URL itself was a "link to" identifying information under the DMCA's catch-all category. The Fifth Circuit rejected a categorical bar on all three, while making clear that most real-world URLs will fail in practice, on any of the three theories, for structural reasons.
CMI theory | What the court requires | Where ordinary permalinks fall short |
|---|---|---|
Domain name as copyright owner (§1202(c)(3)) | The domain must clearly correspond to the current copyright owner | Rights get assigned, transferred and consolidated (as Emmerich itself had done); a domain naming the wrong or a prior owner does not convey notice |
URL slug as title (§1202(c)(1)) | The slug must function like a title, the way a Fifth Circuit precedent already allows for PDF filenames | Slugs are truncated, keyword-optimized and frequently do not match the published headline at all |
Copyright scope matching the work | The registration should cover the specific work the URL identifies | Most publishers register newspapers or issues as a group, not individual articles — weakening the link between a given URL and a given copyright |
URL stability | The address should not be "subject to change" by design | URLs change by definition when content moves server, folder or CMS — the court treated this as working against CMI status, not merely as a technical footnote |
"Link to" CMI (§1202(c)(7)) | The URL must point to a separate page that displays CMI (a rights or credits page) | The court read this narrowly: the URL of the content itself cannot be treated as a link to CMI about itself — that would be circular |
Put together, the court left open a path — but a narrow one. A URL is more likely to be treated as CMI when the domain visibly and currently matches the copyright owner, when the underlying registration covers the specific work rather than a bundled group filing, when the address is structurally stable rather than routinely rewritten, and when the URL is built to communicate the work's identity rather than simply to route a request. Few CMS-generated permalinks are built with any of that in mind today.
The objection: this is one circuit, one aggregator, one unfinished case
Three limits are worth stating plainly, because they change what this ruling is actually worth to a publisher today.
It is interlocutory. The Fifth Circuit answered two certified questions of law and sent the case back for fact-finding. No court has yet held that any specific URL — Emmerich's or anyone else's — actually is CMI. The four-factor pattern above is a reading of what the court is likely to require, not a checklist any publisher has passed yet.
It binds one circuit. The ruling governs Texas, Louisiana and Mississippi. The Ninth Circuit's server test remains good law where it was created, and no circuit split has been formally resolved — the Fifth Circuit reached a related but differently reasoned result rather than adopting or rejecting the Ninth Circuit's rule head-on.
It concerns a news aggregator, not a generative AI system. NewsBreak displays and links to existing pages; it does not train on content, summarize it, or generate new text from it. Extending either holding — the transmit requirement or the CMI reading — to an AI answer engine or a crawler-fed retrieval system is a reasoned inference, not a decided fact. The mechanics may transfer more or less cleanly depending on whether the AI system fetches content live (closer to NewsBreak's Framed View) or trains on and reproduces it from a stored copy (a different legal question entirely, and one this ruling does not reach).
What to check before your next AI or aggregator agreement
The ruling gives publishers a concrete, if narrow, audit to run — ideally before signing anything, not after a dispute starts.
Confirm the domain visibly and currently matches your registered copyright owner, especially after any brand consolidation, imprint restructuring or rights transfer.
Check whether your copyright registrations cover individual articles or only bundle content under an issue or monthly group filing — the latter weakens any CMI claim tied to a single URL.
Audit URL stability. A CMS that rewrites slugs on republish, changes folder structure, or generates ephemeral tracking parameters is manufacturing the exact instability the court treated as evidence against CMI status.
Decide, deliberately, whether your permalink structure is built to identify the work — or only to route a request. Those are different design goals, and the court now treats the difference as legally material.
Separately, and independently of the CMI question, revisit what actually happens when an unfamiliar crawler or agent requests your content. Under the transmit requirement, what your server is willing to answer now does more legal work than what your terms of service say.
Where this connects to usage evidence
A permalink audit is not a branding exercise. It is upstream of licensing. A publisher cannot negotiate on the strength of a URL functioning as CMI if the underlying registration, domain structure and stability do not support that claim. This is the same discipline behind machine-readable provenance for reference publishers: authority has to be encoded, not assumed, before it can survive contact with an AI workflow.
It also connects to what a publisher can prove without depending on a platform's cooperation. Whether a URL was altered, removed or replaced under a different domain is exactly the kind of fact a publisher's own logs can establish — the same custody argument behind which AI Search records publishers should keep themselves. A CMI claim without a dated, contemporaneous record of what the original URL was and when it changed is a much weaker claim.
And it connects to measurement. If an AI system can retrieve and display your content live under the transmit requirement without infringing your display right, the permalink — and whether it survives the trip intact — becomes one of the few remaining signals you control. That is the same gap described in retrieved is not cited: visibility inside an AI answer is not the same as a preserved, attributable record of the source.
This is the layer Citations Logic is built for: turning rights, provenance and attribution into records a publisher holds, rather than assumptions embedded in a URL structure nobody audited. That is what AI usage evidence for publishers means applied to the permalink layer specifically.
The permalink question also sits inside a wider legal portfolio. It is a separate issue from what the Ross appeal means for AI training and competing products — CMI concerns identification and attribution at the point of display, not the training question. And once a permalink can support a rights claim, the next question is commercial: how that claim feeds usage evidence in AI display deals.
Frequently asked questions
Can a URL legally be copyright management information?
The Fifth Circuit held that the DMCA does not categorically foreclose URLs from being CMI. But a URL must clearly convey the specific information the statute requires — such as title or copyright owner — and not function merely as a locational address. Whether any particular URL meets that bar is a fact question the courts have not yet resolved.
Does this ruling mean framing or embedding a publisher's content is now safer or riskier?
Narrower for publishers on the display-right theory. The Fifth Circuit replaced the server test with a "transmit requirement": a site that only requests and displays content actually delivered by the copyright owner's own server is not the one "transmitting" it. A site that blocks or restricts unauthorized requests keeps stronger display-right protection than one that answers every request.
Does this case involve generative AI?
No. Particle Media's NewsBreak is a news aggregation app that links to and frames existing pages; it does not train models or generate summaries. Applying either holding to an AI answer engine or crawler-based retrieval system is a reasoned extension, not something this court decided.
Is this ruling final?
No. It is an interlocutory appeal answering two certified questions of law. The case is remanded to the Southern District of Mississippi for fact-specific findings on whether Emmerich's URLs actually qualify as CMI and whether any removal was intentional.
What should a publisher do now?
Run a permalink audit before the next AI or aggregator agreement: confirm the domain matches the current copyright owner, check whether copyright registrations cover individual works or only group filings, test URL stability over time, and decide deliberately whether the permalink structure is built to identify the work rather than only to route a request.
Continue the evidence chain
AI Search Controls: What Publishers Must Record Themselves
Retrieved Is Not Cited: Why AI Visibility Fails Publishers
Machine-Readable Provenance for Reference Publishers
AI Training, Competing Products, and Copyright: What the Ross Appeal Means for Publishers
AI Display Deals Need Usage Evidence: The Publisher Meter
AI Usage Evidence for Publishers
Book a permalink and usage evidence review
Sources
United States Court of Appeals for the Fifth Circuit — Emmerich Newspapers, Inc. v. Particle Media, Inc., No. 25-60550, opinion filed August 27, 2026
Read the opinion
Justia — Case summary, Emmerich Nwspr v. Particle Media, No. 25-60550 (5th Cir. 2026)
Emmerich Nwspr v. Particle Media case summary
Courthouse News Service — "Fifth Circuit sides with news aggregator in copyright battle"
Fifth Circuit sides with news aggregator in copyright battle
Plagiarism Today — "Server Test Out, Transmission Test In"
Server Test Out, Transmission Test In
Electronic Frontier Foundation — Case page, Emmerich Newspapers v. Particle Media
Emmerich Newspapers v. Particle Media