Photo Licensing Service Qualifies for DMCA Online Safe Harbor–Steinmetz v. ShutterStock

This is a 512(c) online copyright safe harbor case. We rarely see opinions like this any more. In 2022, I’ve blogged just one other 512(c) case (Davis v. Pinterest). (Business Casual v. YouTube should have been a 512(c) case, but the court ruled on other grounds). This time, the defense gets the safe harbor on summary judgment.

The case involves ShutterStock, a photo licensing service. (We used ShutterStock as the blog’s image provider for several years). ShutterStock has a “contributor” program that allows anyone to upload photos into their licensing database in exchange for a fee if licensed. ShutterStock pre-reviews contributor-uploaded photos for image quality, “obvious” copyright infringement, and hate speech/pornography before they are made available for licensing. A contributor uploaded a cropped version of the plaintiff’s registered photo that did not include the photographer’s original metadata. Only 2 visitors saw the subject image in ShutterStock’s database and neither licensed it.

The plaintiff sent a DMCA takedown notice. ShutterStock’s average DMCA turnaround time is 5.5 days, but the court doesn’t say exactly how long it took ShutterStock to respond to this one. Instead, it says that by 40 days later, the URL was down. Five months later, after more complaints from the plaintiff, ShutterStock found “several thumbnails in the edges of its caches servers” and took those down too. ShutterStock uses some third-party agencies that display thumbnails on third-party websites to generate traffic to ShutterStock, and it’s unclear if/when the thumbnails of the subject image were flushed from their caches.

The court holds that ShutterStock qualifies for the 512(c) safe harbor on summary judgment.

Service Provider. (Reminder: the statutory definition of “service provider” is “provider of online services,” so this is not the place for copyright owners to draw their line in the sand. LOL). The court says the “definition is intended to incorporate a broad set of Internet entities, and courts have repeatedly held that defendants offering similar file-sharing-based services qualify as ‘service providers,'” including Photobucket.

To get around this, the plaintiff cited AFP v. Morel. The court is not impressed:

Agency French Presse is an outlier in the DMCA case law. Contrary to the broad reading given by almost every court confronted with similar services and circumstances, there the district court denied summary judgment to Getty Images…

Agence French Presse is not well-reasoned and thus does not constitute persuasive authority. The supposed limitation on Safe Harbor eligibility was grafted on by the district court, which reasoned that Congress must have intended the service provider language to have some limitation. It finds not [sic] support in the text of the statute. In addition, the district court’s decision to draw a line at a profit-driven service provider stands at odds even with the cases it cited.

The court also distinguishes AFP on its facts: “Getty obtained images from Agence French Presse and the right to license the images itself; Getty additionally paid royalties to Agence French Presse in connection with licenses sold. Getty’s employees also actively engaged in licensing the content at issue.” In contrast…

Defendant’s conduct was primarily passive—facilitating rather than performing the licensing. It is undisputed that the Contributor in India uploaded the image. Defendant did not enter into any specific agreements with the Contributor, nor obtain exclusive rights, or pay royalties. And Defendant’s limited role of reviewing uploaded images for objectionable content does not take Defendant beyond the realm of mere facilitator. Although prior to the takedown notice, the image could have been licensed, the undisputed evidence shows that it was not; in fact, there is no evidence that Defendant took any volitional action with respect to Plaintiff’s image or in any way profited from having the image in its portfolio. To wit, whereas Getty was found to have licensed the images in issue to multiple parties, here, the metadata for the image’s URL shows that between its uploading and takedown, the URL was accessed a total of four times by two unique users and generated no revenue whatsoever—in other words, no user or subscriber to Defendant’s stock portfolio licensed or received a copy of the image.

I can see future problems from the court’s approach. First, I don’t think other courts will turn the 512(c) analysis on whether there has been an actual for-pay license of the material or not. Instead, courts will point out that ShutterStock wanted to make money from the image, even if it didn’t succeed. Second, the active/passive distinction has been a recipe for judicial mischief, as we have seen in the 230 context. Here, the court seeks to determine who assembled the database: did the defendants’ employees actively seeking out images, or was the defendant open to all uploaders? While that distinction might help this time, it is more likely to confuse future judges.

Repeat Infringers. “Defendant also has implemented a policy for repeat infringers that complies with the DMCA. The policy is set forth in the CTOS and puts contributors on notice that Defendant retains the right to limit access or terminate accounts for copyright infringement. Defendant also reasonably enforces its policy by enabling copyright owners to submit takedown notices and responding to such notices. Defendant keeps records to identify repeat infringers and regularly terminates contributors’ accounts due to infringing activity, including in this case.”

Standard Technical Measures. The plaintiff claimed that ShutterStock violated this prong by stripping the photo’s metadata. However, the evidence did not indicate that ShutterStock ever had the photo with his metadata. (Metadata also isn’t an STM, but the court doesn’t need to address that).

Stored at User’s Direction. ShutterStock’s prescreening process doesn’t disqualify the file as being stored at a user’s direction: “given that 200,000 images are added to the platform daily, no reasonable factfinder could conclude that Defendant engages in the type of thorough review and approval process suggested by Plaintiff and which might support a finding that the images were not stored at the direction of contributors.”

Knowledge/Expeditious Removal. Oddly, the court doesn’t address the 40 day window between the takedown notice and  confirmation of the URL’s removal. Instead, the plaintiff complained about the lingering backend copies and copies cached in third-party databases. The DMCA takedown notices never gave a URL for the backend copies, and ShutterStock had no responsibility for flushing the third-party caches.

No Control/Financial Benefit. With respect to control, ShutterStock “did not invite or request the Contributor to upload” and can’t be expected to have caught the photo in its prescreening. It took the appropriate removal steps in response to the takedown notice.

With respect to financial benefit, “the image was viewed a total of four times by two discrete users, There is no evidence that the image was licensed or even requested for license. Its mere existence in Defendant’s portfolio of hundreds of millions of images is an insufficient basis for finding financial benefit.” Compare the Fonovisa and Napster cases, which talked about having infringing works in a database as a “draw” for new customers.

False CMI. “Defendant’s use of a watermark does not constitute false CMI. It identifies Defendant as the source of an image downloaded from its portfolio or platform. Second, Plaintiff cannot establish that Defendant had any scienter, let
alone double scienter. The evidence shows that Defendant incorporates or attaches its watermark to prevent, rather than to induce, enable, facilitate, or conceal, infringement”


The court doesn’t discuss fair use, though that was on my mind with the references to a cropped photo. See, e.g., Harbus v. Manhattan Institute.

With a different judge, this summary judgment ruling could have gone the other way, especially regarding the 40 day window to confirm the removal and the financial benefit prong.

For ShutterStock, it surely would have preferred to win on a motion to dismiss to avoid discovery costs. On the plus side, on summary judgment, ShutterStock could answer many of the factual who/what/where/when/why questions about the image in question. That helped the judge rule with confidence about the facts.

Case citation: Steinmetz v. ShutterStock, Inc., 2022 WL 4342174 (S.D.N.Y. Sept. 19, 2022)