ATHOS OVERSEAS LIMITED CORP. v. YOUTUBE, INC.
No. 23-13156
United States Court of Appeals For the Eleventh Circuit
January 7, 2026
FOR PUBLICATION
JORDAN, Circuit Judge:
This case concerns the application of one of the safe-harbor provisions of the 1998 Digital Millenium Copyright Act,
Athos Overseas Limited, which owns the copyright to many classic Mexican and Latin American films, sued YouTube, Inc., YouTube, LLC, and YouTube’s owner, Google, LLC (collectively “YouTube“), alleging copyright infringement based on the unauthorized posting of its copyrighted material on the YouTube website, an internet platform where users can upload video content for public viewing by visitors of the site. Following discovery, Athos filed a motion for partial summary judgment, and YouTube filed its own motion for summary judgment. The magistrate judge issued a report recommending that Athos’ motion be denied and that YouTube’s motion be granted based on
After review of the record and the parties’ briefs, and with the benefit of oral argument, we affirm. We agree with the district court that on this record YouTube was protected by
In 1998 Congress, foreseeing the significance the internet would play in the coming millennium, enacted the DMCA “to update domestic copyright law for the digital age.” Viacom Int‘l Inc. v. YouTube, Inc., 676 F.3d 19, 26 (2d Cir. 2012). The DMCA was meant to “provide certainty for copyright owners and Internet service providers with respect to copyright infringement liability” by creating protections from copyright liability under certain circumstances and by establishing a regime through which copyright owners could reliably enforce their rights. See S. Rep. No. 105-190, at 2 (1998). The DMCA sought to balance the interests of copyright owners and those of the nation by ensuring “that the efficiency of the Internet [would] continue to improve and that the variety and quality of services on the Internet [would] expand.” Id. See also 4 Melville B. Nimmer & David Nimmer, Nimmer on Copyright § 12B.01[C][1] at 12B-28 (2020) (“To balance these divergent interests, the [DMCA] as enacted embodies disparate forms of protection.“).
Congress achieved this balance in part by establishing specific protections for internet service providers. The DMCA safe-harbor provision at issue here applies to “information residing on systems or networks at direction of users” and reads as follows:
(1) In general.—A service provider shall not be liable for monetary relief, or, except as provided in subsection (j), for injunctive or other equitable relief, for infringement of copyright by reason of the storage at the direction of a user of material that resides on a
system or network controlled or operated by or for the service provider, if the service provider— (A)(i) does not have actual knowledge that the material or an activity using the material on the system or network is infringing;
(ii) in the absence of such actual knowledge, is not aware of facts or circumstances from which infringing activity is apparent; or
(iii) upon obtaining such knowledge or awareness, acts expeditiously to remove, or disable access to, the material;
(B) does not receive a financial benefit directly attributable to the infringing activity, in a case in which the service provider has the right and ability to control such activity; and
(C) upon notification of claimed infringement as described in paragraph (3), responds expeditiously to remove, or disable access to, the material that is claimed to be infringing or to be the subject of infringing activity.
Through
Under this notice-and-takedown regime service providers are at risk of losing the protection of the safe-harbor provision should they fail to adequately respond to a claim of copyright infringement. See
Congress intended the notice-and-takedown regime as “a formalization and refinement of a cooperative process that ha[d] been employed to deal efficiently with network-based copyright infringement.” S. Rep. No. 105-190, at 45. It has been described as “a
To qualify for safe-harbor protection under
The safe-harbor set out in
II
We review a district court’s grant of summary judgment de novo, viewing the facts and making inferences in favor of the nonmoving party. See Smith v. Owens, 848 F.3d 975, 978 (11th Cir. 2017). Summary judgment under Rule 56 is appropriate where no material facts are in genuine dispute and one party is entitled to judgment as a matter of law. See id. An issue of fact is material if it has the potential to affect the resolution of the case under the applicable legal principles. See Harrison v. Culliver, 746 F.3d 1288, 1297–98 (11th Cir. 2014). And a fact is genuinely in dispute if, taking the record as a whole, a rational trier of fact could find in favor of the nonmoving party. See id. at 1298.
III
We start by describing the specific technologies and functionalities that are at issue here—what they are, what they do and don’t do, and how they do it—before we wade into application of
Two suites of technologies and functionalities are at issue here. The first suite encompasses the functionalities surrounding YouTube’s moderation and curation of the content on the YouTube website, which are relevant to whether YouTube has the right and ability to control infringing material. The second suite is made up of YouTube’s video-hashing technology as well as the multiple video-hash-matching technologies housed within its various copyright management tools.
The first suite of functionalities relates to the ability of YouTube to moderate and curate user uploads to its website.
Athos contends that YouTube exerts substantial amounts of “control over the users who upload the infringing material, its users in general and the infringing activity on its platform.” Appellant’s Br. at 48 (citing D.E. 137-3 at 15:1–7, 17:12–25, 79:24–87:25, 119:18–24, 120:24–121:04; D.E. 137-6 at 102:25–103:10, 118:17–120:09, 132:08–134:21, 191:11–12; D.E. 137-7 at 53:08–57:20). After careful review of Athos’ record citations, there are only two that reference YouTube’s ability to control the content on the YouTube website. See D.E. 137-6 at 132:08–134:21 (explaining how a user can use the Prevent Copies functionality of YouTube’s copyright web form to have its system prevent the reupload of exact and near-exact copies of already removed videos); D.E. 137-7 at 53:08–57:03 (discussing YouTube’s ability to suggest videos to users and to autoplay videos after one video has finished, and confirming that YouTube can take down any page on its website as well as set its own policies and procedures for what content can be uploaded and viewed).
These matters are not in dispute. YouTube does not contest that its Prevent Copies functionality exists, or that its algorithmic video curation can suggest and autoplay other videos available on the site, or that it has the general power to remove content from the site.
YouTube currently uses video-hashing technology to create an identifying chain of characters, i.e., a hash, for each video uploaded to its website. The video hashes produced by YouTube’s technology are used to facilitate video-hash-matching functionalities available through various copyright management tools that YouTube offers to users. See D.E. 137-6 at 144:16–145:08 (explaining that there are three different copyright management tools offered by YouTube to users that rely on some form of hash-matching technology: (1) Content ID, (2) the Copyright Match Tool, and (3) the Prevent Copies functionality of its copyright web form).
1
The various hash-matching functionalities are each housed in different copyright management tools, but they all have the common capability of comparing a reference video hash to YouTube’s video-hash database to produce a list of video-hash matches that can then be processed in some other manner. The primary differences between the tools for our purposes are the options available to the user for determining what, if any, next steps should be taken with respect to the list produced by the video-hash-matching process. See D.E. 14-2 (discussing Content ID, which allows the user to select from various options, including automatic removal, various monetization settings, or tracking of flagged matches); D.E. 136-3 at 149:14–150:07 (explaining the Prevent Copies functionality, which can prevent the subsequent upload of exact or near-exact copies of already removed videos if a user checks the applicable
YouTube offers its copyright management tools for use by users and does not operate any of them automatically itself because it is the users, i.e., the copyright owners, who have the best knowledge of, and access to, their copyright information. The users must decide what each tool should do with an identified match, as it is their underlying copyrights that are being protected. In the case of Content ID, the user has broad and precise powers to direct the takedown, tracking, or monetization of identified video-hash matches on a region-by-region basis. Without input from a user regarding which actions to take in which regions, YouTube has no way of administering Content ID on the user’s behalf.
Notably, with respect to YouTube’s hash-matching technology, it is possible that a video-hash match identified by the hash-matching process is not actually a violation of the underlying copyright. This can happen, for example, when the use of the
There is no evidence in the record that YouTube has any tool capable of performing a legal evaluation to determine whether an identified video-hash match is actually infringing on a given copyright. And the record indicates that YouTube seeks to avoid mass removals of legal uses of copyrighted material by its users. As a result, none of YouTube’s copyright management tools operate automatically without substantial input and management by the copyright owners, and the most powerful tools are only available to qualifying users.
YouTube offered Athos the use of its various copyright management tools, including Content ID, but Athos declined the offer because it viewed the required user agreement as “monopolistic and abusive.” Athos chose instead to hire a law firm to manually locate potential instances of infringement and send individualized takedown requests to YouTube. Neither Athos nor its counsel ever used Content ID or the Copyright Match tool to assist in the management of its copyrights on the YouTube website.
2
According to Athos, the second suite of technologies is made up of various functionalities that all fall under the umbrella term Content ID. Athos contends this is a system on the YouTube
For this theory, Athos relies exclusively on its interpretation of the deposition testimony of YouTube’s representatives. But the record taken as a whole (and viewed in Athos’ favor) does not support Athos’ characterization of YouTube’s copyright management tools. Our review of the record shows that Athos’ factual assertions rely primarily on unsustainable readings of the deposition of Mr. Zhu, a Content ID product manager.2
Athos’ other record citations do not allow a reasonable jury to find that “YouTube automatically compares each [video] fingerprint to all content in YouTube, as a business policy.” Appellant’s Br. at 15 (citing D.E. 137-6 at 103:7–10, 118:17–120:08). For instance, in response to questions regarding whether any aspects of the video upload process on the YouTube website involve human interaction by YouTube employees, Mr. Zhu stated: “I don’t think I know—so I’m not personally specifically aware of processes where a human interacts with a video during that process, but that
The record citations that Athos relies on also do not support the factual proposition that “[t]he purpose of comparing fingerprints is ensuring YouTube locates and flags infringing content.” Appellant’s Br. at 16 (citing D.E. 137-6 at 70:17–20, 146:12–147:03). Mr. Zhu’s first cited deposition excerpt directly contradicts the proposition Athos asserts. See D.E. 137-6 at 70:17–20 (“The purpose of the [hash-matching] tool is to show videos that are potential matches and allow the user to decide whether they want to request removal of those videos or do something else.“). The second deposition excerpt does nothing to shore up Athos’ unfounded assertion; when Mr. Zhu was asked whether he knows if the different hash matching copyright management tools YouTube has made
3
In sum, as to curation and moderation technologies, the record reflects agreement between the parties that YouTube can algorithmically suggest and auto-play videos, can remove users and videos from the site, and can set its own guidelines for users and uploaded content on the site.
With respect to the hashing technologies, the record does not support Athos’ factual assertions. YouTube does not run Content ID (or any of its other copyright management tools) automatically or universally for all videos uploaded to the YouTube website. Athos, moreover, has offered no evidence that any of YouTube’s video-hash-matching or other tools have any functionality to perform a legal analysis of whether an identified video-hash match constitutes copyright infringement. In fact, YouTube has presented unrebutted evidence that all of its copyright management tools were designed with user input and management in mind because it would be unfeasible for YouTube to operate the tools by itself without such input. Athos simply has not identified any material facts in dispute.
According to Athos, once it notifies YouTube of its ownership of a given copyright and identifies a single location (e.g., a URL) where the copyrighted material appears on the YouTube website, YouTube then has a video hash sufficient for it to locate additional infringing material on the site. Athos argues that this confluence of technologies generates actual or red flag knowledge of infringement that should result in YouTube being required to remove any match generated by the hash-matching technologies or otherwise risk losing its safe-harbor protection under
Athos challenges the district court’s reliance on Second and Ninth Circuit cases considering equivalent issues—Viacom Int‘l, Inc. v. YouTube, Inc., 676 F.3d 19 (2d Cir. 2012), and UMG Recordings, Inc. v. Shelter Cap. Partners LLC, 718 F.3d 1006 (9th Cir. 2013)—because it contends that the approach in these cases “disproportionately protects service providers, disincentivizes the creation of creative work and diminishes the value of obtaining copyright protection.” Appellant’s Br. at 27. We are not persuaded.
A
The safe-harbor provision set out in
The Second Circuit in Viacom concluded that the removal obligation in
The Ninth Circuit came to the same conclusion in UMG, 718 F.3d at 1022. It held “that merely hosting a category of copyrightable content, such as music videos, with the general knowledge that one’s services could be used to share infringing material, is insufficient to meet” either the actual or red flag knowledge
1
Under Viacom and UMG, YouTube is entitled to safe-harbor protection under
As set out below, we have two questions about Nimmer’s interpretation of the red-flag-knowledge provision but ultimately conclude that we need not resolve them to decide this case. Nevertheless, to provide a full explanation we set out our thoughts about the Nimmer view.
First, we think it noteworthy that the examples Nimmer uses specifically concern an internet service provider offering directory services, i.e., services aggregating an array of links to various other websites for their users to more easily locate and visit
Second, the language of
For a video hosting service like YouTube, if general knowledge that the service is being used to occasionally achieve infringement were sufficient to require expeditious removal of all
Importantly for this case, Nimmer also notes that “[e]ven if [a service provider] receives tens of thousands [of] notifications as to particulars within [a website hosted on the service], all it must
2
Athos suggests that YouTube’s video-hash-matching technology was “built to provide actual knowledge of infringement[.]” Appellant’s Br. at 18. But Athos has not advanced any evidence that YouTube has actual knowledge of any infringing material that it does not expeditiously remove. Indeed, Athos appears to concede that YouTube only gains actual knowledge of specific infringement once it receives a takedown request identifying the exact location of particular infringing material. See Appellant’s Br. at 18 (“YouTube does not remove the clips once it becomes aware of the red flag from which infringing activity is apparent; but rather, waits until it receives ‘actual knowledge’ in the form of another
The record reflects that the copyright management tools could not have provided YouTube with actual knowledge because they were never used in the manner Athos suggests. As discussed earlier, Athos has not presented evidence that any of the video-hash-matching tools it argues were designed to provide actual knowledge are operated automatically or universally by YouTube. Even if it had made such a showing, the video-hash matches generated by YouTube’s technologies, as we understand them from this record, cannot give YouTube actual knowledge of infringing material because the tools do not perform any analysis to determine whether the hash matches they generate are legally infringing or not.3
Without any kind of additional analysis being performed by the copyright management tools or YouTube employees, YouTube cannot have “actually or ‘subjectively’ [known] of specific infringement,” because the tools at most produce lists of
3
Athos maintains that, because of its understanding of YouTube’s technologies, a takedown request regarding specific infringement of a given work inherently necessitates removal of multiple other instances of that work appearing in any non-noticed videos on the YouTube platform. See Oral Argument Audio at 12:00–13:47. This is not a plausible or workable reading of the DMCA. Nor, as discussed earlier, is this theory rooted in a correct understanding of how the technologies and functionalities at issue actually work. Again, YouTube does not operate any of its copyright management tools universally or automatically in the manner Athos asserts.
4
Athos’ final argument with respect to knowledge relates to willful blindness. Athos contends that, because of its technologies and functionalities, YouTube behaves in a way that amounts to intentional avoidance of actual or red flag knowledge of infringing material on the YouTube website. This argument is also unconvincing.
“[W]illful blindness is a form of constructive knowledge for contributory trademark infringement.” Luxottica Grp., S.p.A. v. Airport Mini Mall, LLC, 932 F.3d 1303, 1313 (11th Cir. 2019). “Willful blindness occurs when a person suspects wrongdoing and deliberately fail[s] to investigate.” Id. (internal quotation marks and citation omitted). We agree with the Second Circuit that, because its
Though a service provider “would not qualify for the safe harbor if it had turned a blind eye to ‘red flags’ of obvious infringement,” S. Rep. No. 105-190, at 48, YouTube did not act this way here. Congress’ focus on the obviously “infringing nature of” material that service providers cannot be willfully blind to reflects the intended “common-sense result” that service providers “not be required to make discriminating judgments about potential copyright infringement.” Id. at 48–49.
Like the Second Circuit, “we can see no reason to construe [
B
We conclude with the alleged right and ability of YouTube to control infringing material it purportedly derives a direct financial benefit from. Agreeing with the Second Circuit’s consideration
The right and ability to control “requires ‘something more’ than the mere ability to remove or block access to materials on [the service provider’s] website.” Vimeo II, 125 F.4th at 423. The service provider must exert “‘substantial influence’ on user activities” or otherwise have “induced the infringing activity.” Id. at 423–24.
In Vimeo II, the Second Circuit explained that “[c]alling attention to selected videos by giving them a sign of approval or displaying them on a Staff Picks channel (or the contrary, by demoting them) did not restrict the freedom of users to post whatever videos they wished.” Id. at 425. It also concluded that, given the “huge number of videos posted by users on Vimeo,” the plaintiffs there had failed to show that Vimeo staff had intervened in more than a “tiny percentage.” Id. In its view, “denial of eligibility for the safe harbor based on such noncoercive exercises of control over only a small percentage of postings would undermine, rather than carry out, Congress‘[ ] purposes in establishing the safe harbor.” Id. at 426.
We agree with the Second Circuit, and conclude that the same reasoning applies here, where the record reflects only that YouTube can remove material and users from its website, can promote and auto-play select videos algorithmically, and can set
None of the identified “noncoercive exercises of control” on the part of YouTube amount to substantial influence over user activity sufficient to establish it had the right and ability to control any infringing material at issue in this case. See id. “To interpret this provision as [Athos] argue[s]—to deny [YouTube] access to the safe harbor merely because of the tiny influences it exercised—would subject [YouTube] to a huge expense in monitoring millions of posts to protect itself against the possibility of liability for infringements.” Id. at 426–27. Athos’ approach would have us upset the balance between copyright owners and service providers struck by Congress in the DMCA. If that balance is to be recalibrated, it is a matter for Congress.
IV
Aside from specific takedown requests, YouTube’s technologies and functionalities do not produce actual or red flag knowledge of specific infringing material. YouTube, moreover, does not exercise the right and ability to control any of the material on its site for purposes of the DMCA. As a result, YouTube is entitled to
AFFIRMED.