Athos Overseas Limited Corp. v. YouTube, Inc.

UNITED STATES DISTRICT COURT

SOUTHERN DISTRICT OF FLORIDA

Case No. 21-21698-Civ-GAYLES/TORRES

ATHOS OVERSEAS LIMITED CORP.,

Plaintiff,

v.

YOUTUBE, INC., YOUTUBE, LLC, and

GOOGLE, LLC,

Defendants.

___________________________________________/

AMENDED1 REPORT AND RECOMMENDATION

ON THE PARTIES’ MOTIONS FOR SUMMARY JUDGMENT

This matter is before the Court on Athos Overseas Limited Corp.’s (“Athos” or

the “Plaintiff”) Motion for Partial Summary Judgment (“Plaintiff’s Motion”)

[D.E. 125], and Defendants YouTube, Inc., YouTube, LLC, and Google LLC’s

(collectively, “Defendants”) Cross Motion for Summary Judgment (“Defendants’

Motion”) [D.E. 134-1]. The parties filed timely responses [D.E. 139, 138] and replies

[D.E. 147, 150] to each motion, therefore, the motions are now ripe for disposition.2

1 This Amended R&R is strictly limited to correcting minor scrivener’s errors that

were inadvertently present in the original R&R filed on May 16, 2023 [D.E. 171].

Therefore, this Amended R&R does not alter the substance, analysis, or conclusions

of our prior R&R, nor does it affect in any way the procedural or substantive deadlines

that were set in motion by that ruling.

2 On December 30, 2022, the Honorable Darrin P. Gayles referred this case to the

undersigned for a ruling on all pre-trial, non-dispositive matters and a report and

recommendation on all dispositive matters. [D.E. 155].

After careful consideration of the briefing materials, the evidence of record, the

relevant authorities, and for the reasons discussed below, we hereby RECOMMEND

that Defendants’ Motion be GRANTED, Plaintiff’s Motion be DENIED, and the case

be dismissed with prejudice.

I. BACKGROUND3

This lawsuit alleges copyright infringement arising from the posting of

copyrighted materials on YouTube’s website by third-party users. The Plaintiff in

this action is a Panamanian entity owned by Carlos Vasallo (“Mr. Vasallo”), a

businessman and video producer, who holds title to hundreds of classic Mexican films

that have been allegedly uploaded and displayed on YouTube without his prior

authorization.

According to Plaintiff, Defendants are liable under direct and secondary

infringement theories for YouTube’s failure to prevent the systematic re-posting of

Plaintiff’s copyrighted movies to its platform. Plaintiff contends that YouTube has

turned a blind eye to rampant infringement of Athos’ copyrights by refusing to employ

proprietary video-detection software to block or remove from its website potentially

infringing clips, and not just clips specifically identified by URL in Plaintiff’s DMCA

takedown notices. In essence, Plaintiff argues that evidence of YouTube’s advanced

video detection software, in conjunction with the thousands of takedown notices

3 The relevant undisputed facts are taken from Plaintiff’s Statements of Material

Facts in Support of its Motion for Partial Summary Judgment and Response in

Opposition [D.E. 114, 137], and Defendants’ Statements of Fact in Support for their

Motion for Summary Judgment [D.E. 134-2, 151].

Athos has tendered upon YouTube, give rise to genuine issues of fact as to whether

Defendants have forfeited the DMCA’s safe harbor protections.

Defendants, on the other hand, assert that Athos cannot displace YouTube

from the safe harbor of the DMCA because Plaintiff’s construction of the statute is at

odds with its plain language and applicable case law. Defendants submit that

Plaintiff is incapable of meeting its burden of proof insofar as Plaintiff proffers no

evidence of knowledge of specific infringement, relies on a conception of knowledge

that clashes with the DMCA’s non-monitoring provisions, and cannot point to

evidence linking its theory of infringement to any of the particular clips-in-suit. In

light of this evidentiary deficit, Defendants posit, this Court should find that

YouTube’s actions comport with the DMCA and grant judgment on their safe harbor

defense.

A. The Digital Millennium Copyright Act (“DMCA”)

“The DMCA was enacted in 1998 to implement the World Intellectual Property

Organization Copyright Treaty” and to “update domestic copyright law for the digital

age.” Viacom Int'l, Inc. v. YouTube, Inc., 676 F.3d 19, 26–27 (2d Cir. 2012) (quotes

and citations omitted). In passing the DMCA, Congress attempted to strike a balance

between the protection of copyrights and the continued development of the internet.

See Capitol Recs., LLC v. Vimeo, LLC, 826 F.3d 78, 82 (2d Cir. 2016). To that end,

the DMCA enhanced “copyright protection by establishing a ‘notice-and-takedown

regime’ that requires service providers to ‘expeditiously . . . remove . . . material that

is claimed to be infringing,” while expressly relieving “[internet service providers] of

any obligation to monitor the postings of users to detect infringements as a condition

of qualifying for the safe harbor.” Capitol Recs., LLC v. Vimeo, LLC, No. 09-CV-10101

(RA), 2021 WL 2181252, at *2 (S.D.N.Y. May 28, 2021) (quotes and citations omitted)

(alterations in original).

Indeed, the DMCA contains several “safe harbor” provisions that shield

internet service providers (“ISPs”) from liability if they meet applicable statutory

conditions. See Ventura Content, Ltd. v. Motherless, Inc., 885 F.3d 597, 602 (9th Cir.

2018). First, a party seeking the protection of the DMCA must satisfy certain

eligibility criteria, “including the adoption and reasonable implementation of a

‘repeat infringer’ policy that ‘provides for the termination in appropriate

circumstances of subscribers and account holders of the service provider’s system or

network.’” Viacom, 676 F.3d at 27 (quoting 17 U.S.C. § 512(i)(1)(A)).4 In addition to

these threshold criteria, the ISP must also satisfy the requirements of the applicable

safe harbor. The relevant safe harbor provision here is § 512(c), which provides as

follows:

(1) In general.–A service provider shall not be liable . . . 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

4 The statute lists a number of other threshold criteria that are not contested in this

case. See 17 U.S.C. § 512(c).

(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.

17 U.S.C. § 512(c)(1).

Thus, in order to qualify for the protection of this safe harbor, the ISP must

not know of the infringement; the infringement must not be apparent; it must remove

infringing material when it learns about it or upon receipt of a valid DMCA notice;

and it must not directly benefit financially from the infringement in situations when

it can control the activity. See Motherless, Inc., 885 F.3d at 604; Viacom, 676 F.3d at

28.

B. The Alleged Infringement

The relevant facts in this dispute date back to 2014. After finding out that

some of his movies were being displayed on YouTube’s website without his

authorization, Mr. Vasallo retained Gibney Anthony & Flaherty (“Gibney”) in

November 2014 to keep track of clips on YouTube that infringed on Athos’ copyrights.

In addition to this monitoring function, Gibney was tasked with the enforcement of

Athos’ copyrights, which it carried out by compiling and sending DMCA takedown

notices to YouTube on Athos’ behalf.

Following the engagement of Gibney as copyright enforcement agent,

Mr. Vasallo met with YouTube employees in January 2015 to discuss ways in which

YouTube and Athos could cooperate to facilitate Athos’ enforcement efforts. As part

of these discussions, Defendants offered Mr. Vasallo access to YouTube’s most

powerful copyright protection software, Content ID. These talks reached an impasse,

however, after Mr. Vasallo refused to agree to YouTube’s Content Hosting Services

Agreement (“CHSA”), due, in part, to the fact that the CHSA contained a release of

claims provision.5 The CHSA was a standard agreement to which all Content ID

participants had to agree. Unwilling to partake in YouTube’s automated copyright

protection software tools, Athos was relegated to the individualized takedown process

prescribed by the DMCA.

Gibney continued to monitor and tender takedown notices on behalf of Athos,

and, between June and November 2015, it sent a number of demand letters asking

YouTube to take affirmative steps to prevent the re-posting of clips that infringed

upon Athos’ copyrights. Gibney noted in these letters that despite sending hundreds

of DMCA takedown notices and having YouTube remove the noticed videos,

third-party users continued to upload new clips that infringed on Plaintiff’s works (in

many instances, the same works). The letters also claimed that YouTube had the

tools to proactively filter out unauthorized clips—including info-graphic software—

5 [D.E. 100 ¶ 19] (noting that one caveat to the solutions proposed by YouTube was

that “Mr. Vasallo would have to agree to release YouTube from all possible claims”).

During one of his depositions Mr. Vasallo stated that the “terms in the Content

Hosting Agreement are those of a monopoly, and abusive.” [D.E. 134-2 at 281].

that it could use to police its systems and prevent the future posting of infringing

clips without additional input from Athos. [D.E. 134-2 at 109–114].

YouTube responded to Gibney in December 2015 via email. In its response,

YouTube pointed out that it could not operate its copyright protection tools on behalf

of Athos, or any other copyright owner, because only copyright holders know the full

extent of their copyrights. Additionally, YouTube noted that it was its normal

procedure to implement video-matching technology to make sure that identical copies

of noticed and removed clips are never reuploaded; that YouTube had already offered

Athos access to its most powerful copyright protection software; and that Athos had

chosen to forego participation in the Content ID program. Id. at 117. YouTube

reiterated that, if it chose to, Athos had the option of using the automated copyright

tools offered by the platform at no cost. Id.

Almost six years and thousands of DMCA takedown notices later, Athos filed

this lawsuit against Defendants on May 3, 2021, asserting copyright infringement

and related claims based on the public performance, display, and reproduction of

infringing clips on YouTube’s website.6 On March 29, 2022, this Court granted in

part Defendants’ motion to dismiss, dismissing Plaintiff’s FDUTPA and Sherman Act

6 Direct copyright infringement by public performance (Counts I and VIII); direct

copyright infringement by public display (Counts II and IX); direct copyright

infringement by reproduction (Counts III and X); inducement of copyright

infringement (Count IV); contributory copyright infringement (Count V); vicarious

copyright infringement (Count VI); removal and alteration of copyright information

along with intentional distribution of materials lacking copyright information

(Counts VII and XI); Florida Deceptive and Unfair Trade Practices Act (“FDUTPA”)

(Count XII); and violation of § 1 of the Sherman Act for illegal tying (Count XIII).

claims, as well as all copyright claims that accrued prior to May 3, 2018. [D.E. 44 at

10]. On August 16, 2021, Plaintiff voluntarily dismissed its copyright management

information claims (Counts VII and XI). [D.E. 80]. Thus, the only claims currently

before the Court are Plaintiff’s claims for direct and secondary copyright infringement

against Defendants. These claims are premised on approximately 7,000 allegedly

infringing clips that Athos claims displayed segments of roughly 570 of its

copyrighted movies.

C. YouTube’s Copyright Protection Tools

YouTube operates a video-sharing website and mobile application onto which

users may upload video files free of charge. Uploaded files are copied and formatted

by YouTube’s automated computer systems, and then made available for viewing on

YouTube. As of February 2020, more than 500 hours of new video-viewing time was

uploaded to YouTube every minute. [D.E. 134-2 at 416].

YouTube offers an online form (the “webform”) that enables copyright holders

who have a YouTube account to request that YouTube remove allegedly infringing

material from the service. Id. at 3.7 YouTube also accepts DMCA takedown requests

from non-users, such as Athos, by email, fax, and postal mail, and, upon receipt of a

DMCA complaint request, removes the noticed material expeditiously. Id. at 418.

Prior to removal, however, YouTube notifies the uploading user about the removal

request against it and affords it an opportunity to file a counter notification. Id. This

7 In order to properly submit a DMCA notice via the webform, a user must type

certain required information into the form, including his name, the specific URL of

any allegedly infringing clip, and a description of his copyrights.

process reduces the enforcement of invalid or abusive takedown request containing

false or inaccurate assertions of copyright infringement. [Id. at 419–20; D.E. 137-7,

151:2–20].

YouTube also implements a policy that provides for the termination of

subscribers and account holders who are repeat infringers. Id. If a video is removed

based on a DMCA notice, YouTube assigns a strike to the user’s channel.8 YouTube

tracks the number of notices that are processed for each channel, and records strikes

for the removal of videos pursuant to DMCA notices. Users that accrue three or more

copyright strikes are subject to termination for repeat infringement. Upon

termination for repeat infringement, all content and channels associated with the

subject user are removed from YouTube. [D.E. 137-7, 193:20–24]. Youtube has

terminated 1,240 accounts against whose videos Plaintiff submitted takedown

requests. [D.E. 134-2 at 472].

The record also reflects that YouTube has invested hundreds of millions of

dollars in the development of proprietary software aimed at enhancing copyright

protection in its platform. [D.E. 134-2 at 420]. YouTube refers to these features as

its copyright management tools, which consist of the Copyright Webform, Copyright

Match Tool, Content Verification Program, and Content ID. YouTube account

holders have access to these tools free of charge. Id. at 415, 420.

8 A YouTube channel is the space where an account holder stores his videos. It is the

equivalent of a virtual library, in that its primary purpose is to store a user’s content.

[D.E. 137-7, 73:16–25].

The Copyright Match Tool is an automated software that, through the use of

encoding and fingerprinting technology, detects videos that match a user’s

copyrighted material and presents those matches to the user. The user is then

responsible for conducting an assessment concerning the potential infringing nature

of the matches. [D.E. 137, 70:22–72:23]. The user can decide whether to ignore the

match, or file a webform request for its removal from YouTube.9 These scanning and

matching functions are automated and take place without the intervention of any

YouTube employees. [Id. at 118:17–120:5, 163:21–164:3]. In order to use this

software, a right holder must create a YouTube account from which he can review

Match Tool results and make sure that the matches are indeed copyright infringers.

[Id. 73:5–75:19]. These scanning systems are not perfect, and different variables can

affect the accuracy of the software. Id.

YouTube also offers Content ID. Content ID is a highly automated software

available for users with the most complex copyright management needs. [D.E. 134-2

at 456]. Content ID uses similar matching technology as the Match Tool, but unlike

the Match Tool, requires qualifying participants to deliver to YouTube additional

reference files about their copyrighted works, including audiovisual samples,

metadata, and rights ownership information about the protected works. [D.E. 137 at

79:10–81:2, 86:24–88:25, 140:3–141:4]. The audiovisual content in a reference file is

9 Starting in October 2021, YouTube introduced a “prevent copies” option that is

available for anyone filing a takedown request via the webform. This technology uses

hashing and video fingerprinting data to automatically prevent the future upload of

near-identical copies of videos removed pursuant to a valid DMCA takedown request.

[D.E. 137-7 at 132:8–163:24].

not displayed publicly on YouTube unless the uploading user chooses to make that

content public (which is not required). [D.E. 134-2 at 421]. When Content ID finds a

video that matches a reference file, it automatically follows the copyright owner’s

predefined directives for what action to take with such video. Copyright owners may

instruct Content ID to “block” the video from appearing on YouTube, “monetize” the

video by sharing in advertising revenue it generates with YouTube, or simply “track”

views of the video. [Id. at 420–421]. Content ID users have the option, but are not

required, to display or monetize their content on YouTube’s service. [Id.].

II. APPLICABLE PRINCIPLES AND LAW

The court shall grant summary judgment if the movant shows that there is no

genuine dispute as to any material fact and the movant is entitled to judgment as a

matter of law:

A party asserting that a fact cannot be or is genuinely disputed must

support the assertion by: (A) citing to particular parts of materials in

the record, including depositions, documents, electronically stored

information, affidavits or declarations, stipulations (including those

made for purposes of the motion only), admissions, interrogatory

answers, or other materials; or (B) showing that materials cited do not

establish the absence or presence of a genuine dispute, or that an

adverse party cannot produce admissible evidence to support the fact.

Fed. R. Civ. P. 56(c)(1). On summary judgment, the inferences to be drawn from the

underlying facts must be viewed in the light most favorable to the party opposing the

motion. See Matsushita Electric Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 597

(1986) (quoting another source).

At the summary judgment stage, the Court’s function is not to “weigh the

evidence and determine the truth of the matter but to determine whether there is a

genuine issue for trial.” Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 243 (1986). In

making this determination, the Court must decide which issues are material. A

material fact is one that might affect the outcome of the case. See id. at 248 (“Only

disputes over facts that might affect the outcome of the suit under the governing law

will properly preclude the entry of summary judgment. Factual disputes that are

irrelevant or unnecessary will not be counted.”). “Summary judgment will not lie if

the dispute about a material fact is genuine, that is, if the evidence is such that a

reasonable jury could return a verdict for the nonmoving party.” Id.

III. ANALYSIS

The parties have competing views as to whether Defendants are shielded from

infringement liability by operation of § 512(c)’s safe harbor provision. Plaintiff

submits that YouTube has forfeited the liability protections afforded by § 512(c)

because the ISP had knowledge of infringing clips which it failed to timely remove

from its website, in breach of §512(c)(1)(A). The gravamen of Plaintiff’s claim is that,

because YouTube has automated software that scans videos to help users identify

potentially infringing clips, Plaintiff’s DMCA notices imputed on YouTube knowledge

of each and every single clip that infringed on a noticed film. [D.E. 138 at 6–7].

Accordingly, Plaintiff argues that by limiting itself to removing only the clip

specifically identified by URL in the DMCA notice, as opposed to every other clip that

infringed on the relevant movie, YouTube breached its DMCA removal duties. [Id.].

YouTube asserts that Athos’ construction of the DMCA is at odds with both the

statute’s plain language and a long line of applicable case law rejecting such reading.

Specifically, Defendants point out that Athos’ record is devoid of evidence suggesting

knowledge, actual or otherwise, of any non-noticed infringing clips, and that

Plaintiff’s expansive conception of knowledge—i.e., one that imputes specific

awareness on the ISP based on the capabilities of its video surveillance software—

clashes head-on with the DMCA’s non-monitoring provisions. [D.E. 134-4, at 14–17;

150 at 4]. Additionally, Defendants argue that Plaintiff’s case is fatally flawed in that

Athos cannot point to any evidence linking its theory of infringement to any of the

particular clips-in-suit.

For the reasons that follow, we agree with Defendants that all of Athos

arguments are unavailing, and that the available record does not give rise to any

triable issues of fact as to whether YouTube is entitled to the safe harbor protections

of § 512(c).

A. Actual or Red-Flag Knowledge

Although the Eleventh Circuit has yet to interpret the knowledge provisions of

the § 512(c) safe harbor, both the Second and Ninth Circuit have clearly established

that disqualifying knowledge in the context of § 512(c) requires “knowledge of specific

and identifiable infringements of particular individual items.” Viacom Int'l Inc. v.

YouTube, Inc., 718 F. Supp. 2d 514 (S.D.N.Y. 2010), vacated in part on other grounds,

676 F.3d 19 (2d Cir. 2012); see also UMG Recordings, Inc. v. Shelter Cap. Partners

LLC, 718 F.3d 1006, 1022 (9th Cir. 2013) (the DMCA reflects “Congress’ decision to

enact a notice and takedown protocol encouraging copyright holders to identify

specific infringing material to service providers.”). We find the reasoning

underpinning this line of precedent highly persuasive and follow it as our circuit

precedent requires absent a compelling showing otherwise. See Sokol Bros. Furniture

Co. v. Comm'r, 185 F.2d 222, 224 (5th Cir. 1951) (“the decision of another Court of

Appeals upon an issue and principle the same is properly entitled to great weight,

and itself affords persuasive argument in the determination of the present case”).

As noted above, the protections of § 512(c) may not apply to ISPs who “have

actual knowledge” of infringement (“actual knowledge”), or knowledge “of facts or

circumstances from which the infringing activity is apparent” (“red-flag knowledge”).

See Disney Enterprises, Inc. v. Hotfile Corp., No. 11-20427-CIV, 2013 WL 6336286, at

*26 (S.D. Fla. Sept. 20, 2013). “The statutory phrases ‘actual knowledge that the

material . . . is infringing and ‘facts or circumstances from which infringing activity

is apparent’ refer to ‘knowledge of specific and identifiable infringements.’” Viacom,

676 F.3d at 30 (quoting district court) (emphasis added). Thus, to disqualify an ISP

from safe harbor immunity based on red-flag knowledge, “the [ISP] must have

actually known facts that would make the specific infringement claimed objectively

obvious to a reasonable person.” Vimeo, 826 F.3d at 93. The “reasonable person to

whom infringement must be obvious is an ordinary person — not endowed with

specialized knowledge or expertise concerning . . . the laws of copyright.” Id. at 93–

94.

As a threshold matter, we note that the uncontroverted evidence on record

shows that when YouTube became aware of allegedly infringing clips as a result of

Plaintiff’s DMCA takedown requests, YouTube expeditiously removed all of the

noticed items. Although Plaintiff attempts to obscure this fact through allegations

that suggest otherwise, Athos’ vague assertions and half-hearted denials are

completely devoid of any supporting evidence of record and, thus, fail to create an

evidence-based dispute of material fact as to this point. [D.E. 137 ¶¶ 25, 27]; [D.E.

134-2 at 162, 165, 322, 327–328]; [D.E. 100-2].

With that settled, we first turn to the underlying theoretical foundation upon

which Athos builds its case: that YouTube possessed specific knowledge of non-

noticed infringing clips due to the video detection capabilities of its automated

software. [D.E. 138 at 5–6]. According to Plaintiff, given that YouTube employs

automated video fingerprinting software, every time that Athos submitted a DMCA

takedown request regarding one of its movies “YouTube immediately obtained red

flag knowledge of the location of near exact matches of the work identified in the

takedown notice, was willfully blind to that red flag, and failed in its obligation to

expeditiously [] remove, or disable access to, the [infringing] material[.]” Id. at 7

(quotes and citation omitted) (alterations in original). We find this argument

unavailing for multiple reasons.

For starters, courts have repeatedly rejected the theory that an ISP can be

attributed with specific knowledge over non-noticed infringing items by virtue of the

ISP’s video surveillance capabilities. As multiple rulings have put it, requiring ISPs

to use their technologies to identify infringing items out of their own initiative would

be a violation of the DMCA’s non-monitoring and copyright policing principles. See,

e.g., UMG, 718 F.3d 1006 (copyright owners are “better able to efficiently identify

infringing copies than service providers”); Viacom, 676 F.3d at 41 (the DMCA “places

the burden of policing copyright infringements on the copyright owner”); Vimeo, 826

F.3d at 90 (DMCA relieved ISPs “of the obligation to scour matter posted on their

services to ensure against copyright infringement.”).

Indeed, in Viacom the Second Circuit rejected identical arguments to the ones

asserted here by Athos, which were presented in a lawsuit brought by various

television networks against YouTube for the unauthorized display of approximately

79,000 video clips that appeared on the website between 2005 and 2008. Viacom, 676

F.3d at 26. Among other things, the Viacom plaintiffs argued that the manner in

which YouTube employed its automated video identification tools—including limiting

its access to certain users—removed the ISP from the safe harbor. Id. at 40–41. Yet,

the court unequivocally rejected plaintiffs’ arguments, holding that the invocation of

YouTube’s technology as a source of disqualifying knowledge must be assessed in

conjunction with the express mandate of § 512(m) that “provides that safe harbor

protection cannot be conditioned on ‘a service provider monitoring its service or

affirmatively seeking facts indicating infringing activity[.]’”10 Viacom, 676 F.3d at 41

(quoting 17 U.S.C. § 512(m)(1)).

On remand, the district court once again dismissed plaintiffs’ attempt to use

YouTube’s video detection technology as evidence of “knowledge of specific and

identifiable infringements.” See Viacom, 940 F. Supp. 2d at 117. Reiterating

§ 512(m)’s mandate, the district court rejected plaintiffs’ theory that YouTube had

10 Tellingly, Plaintiff’s briefs completely fail to cite to or even mention the directives

imbedded in § 512(m).

been willfully blind to non-noticed infringing clips because “YouTube [could] readily

locate the [non-noticed] infringements by using its own identification tools,” yet failed

to remove these additional clips from its website. Viacom, 940 F. Supp. 2d at 117 n.3.

The court reminded plaintiffs that, under the DMCA, YouTube “had no duty” to use

its automated software to locate and remove infringing clips not identified in the

DMCA takedown notices, because “[u]nder § 512(m), nothing in the applicable section

of the DMCA shall be construed to require YouTube’s ‘affirmatively seeking facts

indicating infringing activity.’” Id. at 117.

Relatedly, the district court also rejected plaintiffs’ objection to YouTube’s

decision to limit access to its video detection software to certain users—an argument

that Athos raises here. According to the Viacom plaintiffs, “YouTube would only use

digital fingerprinting software, which automatically blocks submissions matching

‘reference databases of fingerprints of copyrighted works’ prior to their becoming

available for public view [ ], to filter ‘videos infringing the works of content owners

who had agreed to licensing and revenue sharing deals with YouTube.’” Id. at 120.

Here again, the court relied on the unambiguous text of § 512(m), holding that

“YouTube’s decisions to restrict its monitoring efforts to certain groups of infringing

clips, like its decisions ‘to restrict access to its proprietary search mechanisms,’ do not

exclude it from the safe harbor, regardless of their motivation.” Id. See also Vimeo,

826 F.3d at 98 (“As we made clear in Viacom, § 512(m) relieves the service provider

of obligation to monitor for infringements posted by users on its website. We see no

reason why Vimeo’s voluntary undertaking to monitor videos for infringement of

visual material should deprive it of the statutory privilege not to monitor for

infringement of music.”).

The Second Circuit’s unwillingness to charge an ISP with specific knowledge

of non-noticed infringing items by virtue of the copyright software that the ISP

voluntarily offers to its users has been echoed by the Ninth Circuit. In UMG

Recordings, that Circuit disposed of near-identical claims of actual and constructive

knowledge to those raised by Athos here. In that case, a music publisher sued Veoh,

a video-sharing website, for the unauthorized display of its songs on the platform.

UMG, 718 F.3d at 1011. Like YouTube, Veoh implemented automated software to

prevent copyright infringement, id. at 1012, and like Athos, UMG faulted Veoh for

limiting its removal practices only to the infringing clips specifically noticed on the

relevant takedown requests. Id. And also like Athos does here, the UMG plaintiff

argued that the mere fact that Veoh had the capability and automatically employed

content-detection software amounted to evidence of specific knowledge over non-

noticed clips. See id. at 1012, 1023–1024. In this regard, UMG concluded that “Veoh

should have taken the initiative to use [its] search and indexing tools to locate and

remove from its website any other content by the artists identified in the notices[,]”

as well as information about the “artist, song title and record labels. . . . to find and

remove [non-noticed] unauthorized videos.” Id. at 1025.

The Ninth Circuit, however, disagreed with UMG’s expansive conception of

knowledge, holding that the plain text of § 512(m) and the policy consideration

enshrined in the DMCA foreclosed such a reading. See id. The Ninth Circuit’s ruling

was informed in great part by its perception that, in enacting the DMCA, Congress

understood that “[c]opyright holders know precisely what materials they own, and

are thus better able to efficiently identify infringing copies than service providers like

Veoh, who cannot readily ascertain what material is copyrighted and what is not”:

These considerations are reflected in Congress’ decision to enact a notice

and takedown protocol encouraging copyright holders to identify specific

infringing material to service providers. . . . Congress’ intention is

further reflected in the DMCA’s direct statement that “[n]othing in this

section shall be construed to condition the applicability of subsections

(a) through (d) on ... a service provider monitoring its service or

affirmatively seeking facts indicating infringing activity.” 17 U.S.C.

§ 512(m). Congress made a considered policy determination that the

“DMCA notification procedures [would] place the burden of policing

copyright infringement—identifying the potentially infringing material

and adequately documenting infringement—squarely on the owners of

the copyright.” Perfect 10, Inc. v. CCBill LLC, 488 F.3d 1102, 1113 (9th

Cir.2007). In parsing § 512(c)(3), we have “decline[d] to shift [that]

substantial burden from the copyright owner to the provider.”

Id. at 1121–1022 (alterations in original).

We find this line of cases to be on point, as well as highly persuasive, and

cannot discern a principled basis on this record to depart from their holdings. As the

relevant case law makes clear, evidence of the technologies that ISPs independently

employ to enhance copyright enforcement within their system cannot form the basis

for ascribing disqualifying knowledge of unreported infringing items to the ISP. Such

a conception of knowledge would contradict the plain mandate of § 512(m), “would

eviscerate the required specificity of notice[,] . . . and would put the provider to the

factual search forbidden by § 512(m).” Viacom, 718 F. Supp. 2d at 528. Thus, we find

that Athos’ theory that specific knowledge of non-noticed infringing clips can be

ascribed to Defendants by virtue of YouTube’s copyright management tools fails as a

matter of law.

Athos tries to distinguish this case from both Viacom and UMG, but its attempt

to window-dress its theory of disqualifying knowledge is not compelling. According

to Plaintiff, by asking YouTube to use its technology to remove other infringing clips

besides those specifically itemized in Athos’ DMCA takedown requests, Plaintiff is

not asking YouTube to engage in proactive surveillance or investigative efforts.

[D.E. 138 at 12]. Yet, asking YouTube to go beyond the specific DMCA notices and

effect removals on its own “initiative” (see UMG, 718 F.3d at 1012), whether by means

of “readily available” data or otherwise (see Viacom, 940 F. Supp. 2d at 117, 117 n. 3),

amounts to requiring the ISP to take on an affirmative duty to investigate

infringement. Indeed, Plaintiff’s own record belies its denials. [D.E. 134-2 at 103–

104] (requesting that YouTube use its systems to filter videos based on movie titles

or other key words); id. at 105, 110 (asking that YouTube “proactively remove”

infringing videos by means of its filtering tools)].

In any event, Plaintiff runs headlong against a brick wall erected by the DMCA

that plainly does not require YouTube to do so. See UMG, 718 F.3d at 1023 (9th Cir.

2013) (“[w]e do not place the burden of determining whether [materials] are actually

illegal on a service provider, and [w]e impose no such investigative duties on service

providers.”) (quotes and citation omitted) (alterations in original); Motherless, 885

F.3d at 603 (“The [DMCA] places the burden of policing copyright infringement on

the copyright owner, not on the person or firm storing and hosting the material.”).

Irrespective of how Athos phrases its request, charging YouTube with the

affirmative obligation of going beyond the specific URLs identified in Plaintiff’s

DMCA takedown requests would in effect shift from the copyright owner to the ISP

the burdens of policing and identifying infringement on its systems. This burden

shift would contravene Congress’ calculated choices and the DMCA’s enacted text.

Plaintiff submits nevertheless that “[w]ith the improvements in artificial intelligence,

databases, and other technological advancements, [ISPs] should be required to take

reasonable steps to anticipate and filter potential copyright infringements[,]” [D.E.

100 ¶53], and that “[w]hile it might have been reasonable to limit the removal to

specific urls in an earlier age, the goal of the statute is to make it flexible for the

evolving technology.” [D.E. 138 at 5].

Yet Plaintiff’s plea misses the mark. The question before this court is not what

YouTube or other ISPs should be required to do, but whether YouTube’s acts are

consistent with the statutory scheme set forth by the DMCA as currently enacted.

And while Plaintiff would like for this court to substitute the existing DMCA “notice

and take-down” regime for an amorphous “notice and stay-down” mandate, we cannot

do this just because it makes sense from a copyright holder’s perspective.11 We are

required to enforce the policy choices adopted by Congress rather than recasting the

statute with a critic’s scalpel. See United States v. Garcon, 54 F.4th 1274, 1284 (11th

Cir. 2022) (“Congress often legislates at the macro level, not on a micro scale. One

11 See 4 Nimmer on Copyright § 12B.04[A][2][b] (discussing ongoing debates about

the desirability of reforming the DMCA so that its text reflects a “notice and stay-

down” requirement).

consequence of this approach is that legislation sometimes ‘sweep[s] too broadly’ by

‘affording protection and relief to some who are not truly deserving or aggrieved,’ even

as it sweeps ‘too narrowly’ by ‘failing to reach some who are more deserving or

aggrieved.’ Because this ‘[i]mperfection’ stems from the ‘nature of [the] political

process,’ we consider the rationality of the overall statutory scheme and not whether

‘a particular application of the [scheme] may lead to an arguably anomalous result.’”)

(cleaned up quotations and citations omitted).

B. No Knowledge Over Specific Clips-in-Suit

In addition to running afoul of the DMCA’s specific notice and non-monitoring

provisions, Athos’ case also falls apart from an evidentiary standpoint; that is so,

because even if we accept Athos’ proposition that YouTube’s video-detection software

can be used to impute disqualifying knowledge under § 512(c), Plaintiff’s record is

still entirely devoid of evidence establishing that YouTube acquired knowledge of

infringement relative to any of the particular clips-in-suit here.

A copyright-infringement lawsuit is a claim by “a specific plaintiff against a

specific defendant about specific copyrighted [works]; it is not a lawsuit against

copyright infringement in general.” Perfect 10, Inc. v. Giganews, Inc., 847 F.3d 657,

673 (9th Cir. 2017). In the context of infringing video clips on YouTube’s website, the

Second Circuit has instructed that, “[b]y definition, only the current clips-in-suit at

are issue[,]” and the court must determine “whether any specific infringements of

which YouTube had knowledge or awareness correspond to the clips-in-suit in [the]

action[].” Viacom, 676 F.3d at 34. See also Vimeo, 826 F.3d at 99 (a record that is

devoid of evidence “relat[ing] to any of the videos at issue in this suit,” is “insufficient

to justify a finding of red flag knowledge”).

Here, Athos has failed to present any tangible evidence to establish that, had

YouTube used its video-detection technology as it suggests, the software would have

identified, blocked, or removed any of the specific clips-in-suit in dispute in this case.

This evidentiary deficit is fatal to Athos’ case. See Capitol Recs., LLC v. Vimeo, LLC,

972 F. Supp. 2d 537, 545 (S.D.N.Y. 2013) (“In the absence of evidence showing that a

Vimeo employee watched a particular Video–in–Suit, a reasonable juror could not

conclude that Defendants had knowledge of the video’s contents.”). Instead of

specifying what about its numerically undefined clips-in-suit12 would have made

them discoverable by YouTube’s software, or how YouTube employees interacted with

any of them, Athos conclusorily asserts that its DMCA notices “gave [YouTube] red

flag knowledge of near exact matches of infringing material that, through its

computer-based system, YouTube voluntarily fingerprints and automatically

identifies the material [sic]. Because the material matched other infringing content,

it is obvious to the reasonable person that the matched content is infringing.” [D.E.

138 at 8]. But this “obviousness theory” is both speculative and factually inaccurate.

To begin with, Plaintiff conflates two concepts that are separate and distinct

in the context of YouTube’s copyright protection software: automated video matches

12 Plaintiff’s briefs do not indicate exactly how many of the infringing clips for which

it sent DMCA takedown requests are subject to this lawsuit. In its response in

opposition, Plaintiff simply notes that it had to send more than 3,389 repeated

takedown notices, purportedly between May 3, 2018, and May 3, 2021 (the relevant

timeframe as limited by the applicable statute of limitations).

and actual infringements. As explained by YouTube’s copyright management tools

representative, software-identified video matches are not necessarily tantamount to

copyright infringements. [D.E. 137-7, 74:21–25]. Rather, the software detects code,

audio, or visual cues that may match those of a copyrighted work, and presents those

matches to the owner for inspection. Thus, while YouTube systems may be well

equipped to detect video matches, the software does not necessarily have the capacity

to detect copyright infringements. See id. Further, the accuracy of these

automatically identified matches depends on a wide range of factors and variables.

[Id. at 75:1–10, 108:2–110:17, 113:3–114:25]. That is why users, not YouTube, are

required to make all determinations as to the infringing nature of software selected

matches. [Id.].

Second, Plaintiff does not point to any evidence showing that YouTube,

through its employees, ever came into contact, reviewed, or interacted in any way

with any of the purportedly identified video matches for which Athos was allegedly

required to send subsequent DMCA takedown notices (i.e., the clips-in-suit). As

explained by YouTube’s product manager, the processes of uploading, fingerprinting,

scanning, and identifying video matches is fully automated, involving minimal to no

human interaction in the part of YouTube. [Id. at 68:22–69:18, 118:17–119]. The

record shows that upon upload of a video to YouTube, a chain of algorithmic processes

is triggered, including the automated scanning and matching of potentially

overlapping content. If the software detects potential matches, that list of matches

is automatically directed towards the copyright owner, by being displayed inside the

user’s YouTube interface. [Id. at 68:22–70:25]. Therefore, the record only reflects that

YouTube does not rely on human involvement during this specific phase of the

scanning and matching detection process, and Plaintiff does not proffer any evidence

showing otherwise.

In this respect, Plaintiff’s only evidence related to the specific clips-in-suit is

that those clips were uploaded to YouTube by third-party users at some point between

2018 and 2021, and that by virtue of YouTube’s automated software they were

scanned and fingerprinted. However, nothing in the record suggests that YouTube

employees view every single match identified by the software—let alone every single

video uploaded to Youtube—nor that YouTube employees interacted with any of the

specific clips-in-suit here. Without this evidence, Athos cannot establish red-flag

knowledge. See Vimeo, 972 F. Supp. 2d at 545 (granting summary judgment “with

respect to the ten Videos–in–Suit for which the only evidence of employee interaction

was that ‘Plus’ users had uploaded the videos.”).

Plaintiff’s reliance on Disney and ContextLogic, the only two cases it cites in

support of its argument, is unavailing because the facts in those cases are readily

distinguishable. See Disney Enterprises, Inc. v. Hotfile Corp., 2013 WL 6336286 (S.D.

Fla. Sept. 20, 2013); Venus Fashions, Inc. v. ContextLogic, Inc., 2017 WL 2901695

(M.D. Fla. Jan. 17, 2017). Disney was decided on § 512(c)’s repeat-infringer policy

prong—a requirement that is not in contention here. See Disney, 2013 WL 6336286

at *24. Further, Disney put forth tangible evidence demonstrating that Hotfile

engaged in egregious disregard of known infringements, directly encouraged piracy

in its website, and became attuned the infringing nature of the files within its system.

Id. at *28.

ContextLogic, on the other hand, was decided in the preliminary injunction

context, and involved the use of fingerprinting technology to identify photographs,

not videos. ContextLogic, 2017 WL 2901695 at *2. In addition, the plaintiff

introduced evidence showing that the defendant used infringing items in promotional

and advertising materials it emailed to clients, and further displayed these infringing

items on its sign on page. Id. at *10. Moreover, to the extent that this unpublished

district court case can be read to suggest that ISPs are required to affirmatively use

fingerprinting technology to search for additional instances of non-noticed infringing

content, it erroneously departs from the text of the statute and the many cases, which

deserve far greater weight, that have rejected that idea.

Based on the available record, Plaintiff simply cannot make any factual

showing that YouTube acquired knowledge of specific infringement relative to the

particular clips-in-suit. And without “evidence that would allow a clip-by-clip

assessment of actual [or red-flag] knowledge[,]” Athos cannot withstand summary

judgment. Viacom, 940 F. Supp. 2d at 115–117. See also Vimeo, 826 F.3d at 99

(“because [the] evidence was not shown to relate to any of the videos at issue in this

suit, it is insufficient to justify a finding of red flag knowledge, under the principle of

Viacom, as to those specific videos.”).

Thus, even if Plaintiff’s expansive conception of § 512(c) disqualifying

knowledge were to apply to this case, the lack of evidence regarding the specific clips-

in-suit renders Athos’ claims of safe harbor disqualification untenable.

C. Financial Benefit and Right and Ability to Control

Finally, Athos argues that YouTube does not qualify for safe harbor protections

because it is in breach of § 512(c)(1)(B). “Apart from the foregoing knowledge

provisions, the § 512(c) safe harbor provides that an eligible service provider must

‘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.’”

Viacom, 676 F.3d at 36 (quoting 17 U.S.C. § 512(c)(1)(B)).

Athos claims that YouTube does not meet these requirements because, in its

view, the DMCA codified the common-law doctrine of vicarious-copyright liability,

and under this standard YouTube could be found liable. [D.E. 138 at 14–15]. This

view, however, has been expressly rejected by both the Second and Ninth Circuits as

inconsistent with the text and purpose of o the DMCA. See Viacom, 676 F.3d at 37

(holding that to adopt common law principles of vicarious liability “in the DMCA

context . . . would render the statute internally inconsistent.”); UMG, 718 F.3d at

1029 (“In light of the DMCA’s language, structure, purpose and legislative history,

we are compelled to reject UMG’s argument that” the vicarious liability standard

should be used “to evaluate whether Veoh had sufficient ‘right and ability to control’

infringing activity under § 512(c).

Rather, as the Second and Ninth Circuits have explained, the “right and ability

to control” must “require[ ] something more than the ability to remove or block access

to materials posted on a service provider's website.” Viacom, 676 F.3d at 38.

Otherwise, the ability to remove or block access, which is a “prerequisite to safe

harbor protection under § 512(c)(1)(A)(iii) & (C) would at the same time be a

disqualifier under § 512(c)(1)(B).” Id. at 37. Instead, possession of the “right and

ability to control” contemplates circumstances in which “a service provider exert[s]

substantial influence on the activities of users.” Id. at 38; see Ventura, 885 F.3d at

613 (“To have the right and ability to control, a service provider must be able to exert

‘substantial influence’ on its users’ activities.”).

Based on the evidence in this record, Athos has failed to create a genuine

dispute as to whether YouTube exerted substantial influence over infringing material

on its website. First, Plaintiff’s argument that YouTube’s decision to restrict access

to its video detection software somehow amounts to the right and ability to control

lacks merit. As noted earlier, “YouTube cannot be excluded from the safe harbor by

dint of a decision to restrict access to its proprietary search mechanisms.” Viacom,

676 F.3d at 41. See also Hendrickson v. eBay, Inc., 165 F. Supp. 2d 1082, 1094 (C.D.

Cal. 2001) (“The legislative history shows that Congress did not intend for companies

such as eBay to be penalized when they engage in voluntary efforts to combat piracy

over the Internet.”).

Likewise, Athos’ argument that YouTube’s management of advertising,

monetization, and algorithmic video curation can be construed as control is

unfounded. Indeed, these claims have all been repeatedly rejected by applicable case

law. See, e.g., Viacom, 940 F. Supp. 2d at 120–121 (neither YouTube’s “search

technologies,” nor its “related videos” feature, amounted to “control” under

§ 512(c)(1)(B)); Davis v. Pinterest, Inc., 601 F. Supp. 3d 514, 535 (N.D. Cal. 2022) (no

showing of § 512(c)(1)(B) control based on “Pinterest’s control over its algorithms or

the advertising on its platform”); Vimeo, 972 F. Supp. 2d at 528–530 (same, despite

service’s promotion of particular user-uploaded content for greater visibility,

demotion of undesirable content, and highlighting of popular content).

In the same vein, Athos has made no showing that YouTube received a

financial benefit that was “distinctly attributable to the infringing material at issue.”

Motherless, 885 F.3d at 613. Plaintiff cites no evidence that YouTube encourages

infringement on its website, or that YouTube promotes advertising by relying on

infringing material. See Downs v. Oath Inc., 385 F. Supp. 3d 298, 307 (S.D.N.Y. 2019)

(granting summary judgment on ISP’s § 512(c) safe harbor defense). To the contrary,

YouTube’s uncontested evidence establishes that of the total clips-in-suit, 91%

generated no advertisement revenue on YouTube, that YouTube employs

mechanisms designed to avoid showing ads in connection with content that has been

claimed by a copyright owner, and that if any ad appeared before or after a clip-in-

issue, that ad was merely coincidental. [D.E. 134-2 at 587–89]. See also Downs, 385

F. Supp. at 307 (no financial benefit where site “simply ran advertisements on user-

generated articles, some of which inevitably contained infringing material”).

In sum, having found no genuine issues of material fact, we hold that

Defendants are entitled to judgment as a matter of law on their § 512(c) safe harbor

defense because Athos’ construction of the statute is at odds with its plain language

and applicable case law.

IV. CONCLUSION

For the foregoing reasons, the Court RECOMMENDS that Defendants’

Motion for Summary Judgment on their safe harbor affirmative defense [D.E. 134-1]

be GRANTED, and that Plaintiff’s Motion for Partial Summary Judgment [D.E. 125]

be DENIED. Judgment should be entered for Defendants and the case CLOSED.

DONE AND SUBMITTED in Chambers at Miami, Florida, this 29th day of

August, 2023.

/s/ Edwin G. Torres

EDWIN G. TORRES

United States Magistrate Judge