Second Circuit Limits DMCA § 512(c) Safe Harbor When Platforms Manually and Discretionarily Curate User Uploads

Introduction

McGucken v. Shutterstock, Inc. (2d Cir. Feb. 10, 2026) addresses the scope of two Digital Millennium Copyright Act (DMCA) regimes frequently invoked by online content businesses: (1) the DMCA’s false and removed copyright management information (CMI) provisions, 17 U.S.C. § 1202, and (2) the DMCA’s Online Copyright Infringement Liability Limitation Act (OCILLA) “safe harbor” for user-stored material, 17 U.S.C. § 512(c).

Plaintiff Elliott McGucken, a professional photographer, alleged that hundreds of his photos were uploaded by Shutterstock contributors, displayed for licensing, and in some instances licensed to customers. He sued Shutterstock for copyright infringement (17 U.S.C. § 106) and for false/removed CMI (17 U.S.C. § 1202(a), (b)). The district court granted summary judgment to Shutterstock on all claims based largely on § 512(c) safe harbor and lack of § 1202 scienter. On appeal, the Second Circuit affirmed the § 1202 rulings but vacated and remanded the infringement claims because factual disputes remained as to two key § 512(c) elements: whether infringement occurred “by reason of the storage at the direction of a user,” and whether Shutterstock had the “right and ability to control” the infringing activity.

Summary of the Opinion

  • § 1202 (false/removed CMI): Affirmed summary judgment for Shutterstock. The court found no triable evidence that Shutterstock had the required scienter—knowledge and intent (or “double scienter” under § 1202(b))—for liability based on watermarking and metadata removal.
  • § 512(c) safe harbor (copyright infringement): The court agreed Shutterstock satisfied several prerequisites (service provider status; repeat infringer policy; no interference with standard technical measures; lack of actual/red-flag knowledge; expeditious takedown), but held summary judgment was improper due to factual disputes about:
    1. whether the alleged infringement occurred “by reason of the storage at the direction of a user,” and
    2. whether Shutterstock had the “right and ability to control” the infringing activity.
  • Newly adopted standards:
    • The court adopted the Ninth Circuit’s formulation for “reasonable implementation” of a repeat infringer policy under § 512(i)(1)(A).
    • The court articulated a general rule for when front-end screening can defeat § 512(c)(1)’s “at the direction of a user” requirement: manual, substantive, discretionary review imposing the platform’s aesthetic/editorial/marketing judgment may take the content outside § 512(c).

Analysis

Precedents Cited

1) Second Circuit safe-harbor framework and knowledge standards

The court’s § 512(c) analysis is anchored in Viacom Int'l, Inc. v. YouTube, Inc., which established that safe-harbor eligibility turns on knowledge of “specific instances of infringement” and defined “red flag” knowledge as subjective awareness of facts making infringement objectively obvious. Applying Viacom Int'l, Inc. v. YouTube, Inc., the court rejected attempts to infer knowledge from prior, different disputes involving Shutterstock or contributors, emphasizing that § 512(c)(1)(A)(i) requires awareness of the particular infringements at issue.

The court relied heavily on its recent decision Capitol Records, LLC v. Vimeo, Inc. ("Vimeo II") to describe OCILLA as a statutory “compromise” and to frame how moderation, review, and site design interact with § 512(c)’s “direction of a user” and “right and ability to control” elements. In particular, Capitol Records, LLC v. Vimeo, Inc. ("Vimeo II") supplied the “substantial influence” test for § 512(c)(1)(B) and cautioned that Congress did not intend to disqualify providers for excluding unlawful content or designing a site for certain consumer preferences.

2) What “storage at the direction of a user” can include

To distinguish protected “accessibility-enhancing” functions from disqualifying editorial control, the court cited:

  • Viacom Int'l, Inc. v. YouTube, Inc. (safe harbor can extend to functions facilitating access to user-stored materials; but suggests “manual selection of copyrighted material” may fall outside § 512(c)).
  • BWP Media USA Inc. v. Polyvore, Inc. (safe harbor coverage for certain automated functions; and Judge Walker’s concurrence emphasizing “solely to facilitate access by users” as a touchstone).
  • The court also drew from other circuits to delineate permissible screening: CoStar Grp., Inc. v. LoopNet, Inc. (Fourth Circuit: “cursory” screening does not defeat § 512(c)); BWP Media USA, Inc. v. Clarity Digit. Grp., LLC (Tenth Circuit: screening/automated processes can remain within § 512(c)).

3) Ninth Circuit guidance adopted and integrated

The Second Circuit expressly adopted and operationalized Ninth Circuit approaches in two critical areas:

  • Perfect 10, Inc. v. CCBill LLC (repeat infringer policy): the court adopted the standard that implementation is reasonable if, under “appropriate circumstances,” the provider terminates users who repeatedly or blatantly infringe. Ventura Content, Ltd. v. Motherless, Inc. reinforced that § 512(i) does not require perfection.
  • “Direction of a user” boundary: the court treated Ninth Circuit decisions as a “helpful starting point,” citing Ventura Content, Ltd. v. Motherless, Inc. and Mavrix Photographs, LLC v. LiveJournal, Inc. for the distinction between passive hosting/accessibility enhancement and disqualifying manual/substantive review or editorial judgment. It also cited UMG Recordings, Inc. v. Shelter Cap. Partners LLC for the proposition that active participation in file uploading can defeat § 512(c).

4) “Right and ability to control” and “substantial influence”

The court’s § 512(c)(1)(B) discussion relied on Capitol Records, LLC v. Vimeo, Inc. ("Vimeo II") for the controlling “substantial influence” test and for factors like the “coercive effect and frequency” of platform intrusions into user autonomy. It then connected that framework back to Viacom Int'l, Inc. v. YouTube, Inc., which (citing Perfect 10, Inc. v. Cybernet Ventures, Inc.) recognized that detailed instructions about content, layout, and appearance can evidence control. The opinion also cited Wolk v. Kodak Imaging Network, Inc. for the idea that control may take the form of prescreening and extensive advice or editing.

5) § 1202 scienter (“double scienter”) and automated pipelines

On the CMI claims, the court applied Mango v. BuzzFeed, Inc., which articulates § 1202(b)’s “double-scienter requirement”—knowledge that CMI was removed/altered without authority and knowledge (or reasonable grounds to know) that distribution will “induce, enable, facilitate, or conceal” infringement. The court found the record lacked evidence comparable to Mango v. BuzzFeed, Inc. (where the defendant knew removing/altering attribution would conceal lack of authority).

The court also found persuasive the reasoning of Zuma Press, Inc. v. Getty Images (US), Inc., where an automated process modifying CMI did not show the requisite scienter under § 1202(b).

6) Other cited authorities shaping procedure and record limits

  • Summary-judgment principles: Bey v. City of New York, Schwebel v. Crandall, Bustamante v. KIND, LLC, Souza v. Exotic Island Enters., Inc., Heublein, Inc. v. United States, and Anderson v. Liberty Lobby, Inc. framed burdens and the need for evidence beyond a “scintilla.”
  • Knowledge in the defendant’s possession at summary judgment: the court discussed Friedman v. Live Nation Merch., Inc. (and its reliance on Nissan Fire & Marine Ins. Co. v. Fritz Cos., Inc.) to reject the notion that scienter elements become unfairly unreachable at summary judgment.
  • Limits on appellate record: Amara v. Cigna Corp. supported refusing to consider extra-record articles about metadata.
  • Service-provider status disputes: the court rejected a narrower view from Agence France Press v. Morel and aligned with cases finding marketplaces and hosting platforms to be service providers, including Corbis Corp. v. Amazon.com, Inc., Hendrickson v. eBay, Inc., and Steinmetz v. Shutterstock, Inc..

Legal Reasoning

1) § 1202: scienter is the gatekeeper

The court treated the scienter elements as dispositive. Even if watermarking could theoretically constitute false CMI (the court expressly declined to decide that question), liability under § 1202(a) requires proof that the defendant acted “knowingly and with the intent to induce, enable, facilitate, or conceal infringement.” Likewise, § 1202(b) requires knowledge that CMI was removed/altered without authority and knowledge/reasonable grounds to know distribution will facilitate concealment or infringement.

Shutterstock’s evidence framed its watermark as anti-misuse identification of source/distributor (not authorship), and metadata stripping as motivated by non-infringement concerns (viruses and personally identifiable information). McGucken offered no record evidence of Shutterstock’s knowing falsity, intent to conceal infringement, or awareness that removed metadata was unauthorized as to his works. Without a triable scienter showing, both § 1202 theories failed as a matter of law.

2) § 512(i) threshold holdings, including a newly adopted repeat-infringer standard

The court held Shutterstock satisfied all three threshold safe-harbor prerequisites:

  • Service provider: Shutterstock fits § 512(k)(1)(B)’s broad definition. Licensing user-uploaded images does not categorically exclude service-provider status.
  • Repeat infringer policy: The court adopted Perfect 10, Inc. v. CCBill LLC: implementation is reasonable if, under “appropriate circumstances,” the provider terminates users who repeatedly or blatantly infringe. It also adopted the “not perfection” emphasis consistent with Ventura Content, Ltd. v. Motherless, Inc.. A single missed termination, given the scale of Shutterstock’s platform, could not create a triable dispute.
  • No interference with standard technical measures: McGucken did not prove metadata qualifies as a “standard technical measure” under § 512(i)(2)(A)’s “broad consensus” and standards-process requirements.

3) § 512(c)(1): the opinion’s central clarification—curation versus facilitation

The court held Shutterstock met several § 512(c) elements (no actual/red-flag knowledge; expeditious takedown). But it vacated summary judgment because the record permitted competing characterizations of Shutterstock’s role in what content appears on its marketplace.

The court announced a general rule: when a service provider engages in manual, substantive, and discretionary review of user uploads—imposing its own aesthetic, editorial, or marketing judgment on a case-by-case basis—then the appearance and storage of that material is no longer “at the direction of a user” under § 512(c)(1). By contrast, “rote and mechanical” screening aimed at excluding unlawful material, enforcing terms, or basic category-limiting may remain protected.

Applying that rule, the court found evidence pointing both ways. Shutterstock emphasized quick review time (10–20 seconds) and a high acceptance rate (about 93%). McGucken pointed to Shutterstock’s own statements about “high standards,” extensive rejection reasons (focus, exposure, lighting), guidance to contributors, and deposition testimony suggesting reviewers apply subjective judgment (e.g., distinguishing accidental versus intentional focus and accepting technically imperfect but “unique” images). That evidentiary clash created a jury question: does Shutterstock merely enforce baseline standards to facilitate access to user uploads, or does it curate a commercial image collection through discretionary aesthetic control?

4) § 512(c)(1)(B): “right and ability to control” tied to “substantial influence”

The court treated “right and ability to control” as a fact-intensive inquiry aligned with its newly articulated “curation” concern. Drawing from Capitol Records, LLC v. Vimeo, Inc. ("Vimeo II"), it focused on whether Shutterstock’s practices intruded into user autonomy with sufficient coercive effect and frequency to constitute “substantial influence.”

The same facts relevant to “direction of a user” mattered here: front-end review of every image (a gatekeeping function), extensive submission guidance, and quality-based criteria suggesting more than illegality or category design. Because the district court had resolved § 512(c)(1)(B) solely on the control prong, it did not reach whether Shutterstock receives a “financial benefit directly attributable to the infringing activity,” which may become important on remand if a factfinder finds “right and ability to control.”

Impact

  • Marketplace and stock-media platforms face heightened § 512(c) scrutiny: Platforms that pre-screen every submission for quality and selectively build a “collection” now face a clearer risk that a court will deem the material not stored “at the direction of a user,” undermining § 512(c) at summary judgment.
  • “Curation” is the key liability lever: The opinion supplies plaintiffs a more concrete theory to defeat safe harbor: prove that platform review is manual, substantive, and discretionary—especially where aesthetic/editorial judgment is applied.
  • Second Circuit aligns with (and imports) Ninth Circuit doctrine: By adopting Perfect 10, Inc. v. CCBill LLC for repeat-infringer policy reasonableness and leveraging Mavrix Photographs, LLC v. LiveJournal, Inc./Ventura Content, Ltd. v. Motherless, Inc. on “direction of a user,” the court harmonizes standards across major internet-law jurisdictions, likely increasing uniformity while shifting litigation to fact disputes about platform operations.
  • § 1202 claims remain difficult without specific scienter evidence: The ruling reinforces that automated metadata removal and standardized watermarking, without more, will often fail the scienter requirements—pushing plaintiffs toward discovery targeted at internal knowledge, intent, or specific concealment-related conduct.

Complex Concepts Simplified

DMCA Safe Harbor (§ 512(c))
A liability shield for online service providers when infringement happens because users store content on the provider’s system—so long as the provider meets statutory conditions (policies, lack of knowledge, takedowns, and limits on control/benefit).
“By reason of the storage at the direction of a user”
The idea that the user—not the platform—caused the content to be on the site. This case clarifies that if the platform manually and discretionarily curates which uploads will appear (using aesthetic/editorial/marketing judgment), a jury may find the content is not there “at the direction of a user.”
Actual knowledge vs. “red flag” knowledge
Actual knowledge is awareness of specific infringement. “Red flag” knowledge exists when the provider is subjectively aware of facts that would make the infringement objectively obvious to a reasonable person.
“Right and ability to control” / “substantial influence”
More than general platform control. It refers to meaningful, coercive influence over user activity—such as prescreening coupled with extensive instructions that shape what users can post.
CMI (Copyright Management Information) and “double scienter” (§ 1202(b))
CMI includes attribution and identifying information (often embedded as metadata). “Double scienter” means the plaintiff must show the defendant knew CMI was removed/altered without authority and also knew (or had reason to know) that distributing the work would facilitate or conceal infringement.

Conclusion

McGucken v. Shutterstock, Inc. meaningfully tightens the analytical link between a platform’s front-end content practices and DMCA § 512(c) protection. While reaffirming that § 1202 claims require concrete scienter evidence, the Second Circuit clarified that a platform’s manual, substantive, discretionary screening—particularly when it reflects aesthetic or editorial judgment—can create triable issues as to whether content is stored “at the direction of a user” and whether the platform has the “right and ability to control” infringement. The case signals that, for curated stock-content marketplaces, safe-harbor eligibility may turn less on formal labels (“user-generated”) and more on operational reality: whether the platform behaves like a neutral host or a curator of a commercial collection.