The recent California district court decision dismissing the complaint in X Corp. v. Bright Data Ltd. could have significant implications for companies that rely on their terms of use to prohibit unauthorized “data scraping” — that is, using automated tools to extract data from a website or online services.

While data scraping was the subject of significant jurisprudence in the dot-com era, it has again received attention as a number of companies have asserted that artificial intelligence (AI) model builders scraped websites or online services without permission to extract data to train their models.

In granting Bright Data’s motion to dismiss, the Northern District of California relied on the doctrine of “conflict preemption” and held that claims that data scraping breached X Corp.’s terms of use impermissibly conflicted with the Copyright Act. In the court’s view, issues of data scraping are typically better addressed through copyright law.

Background

Plaintiff X Corp. owns and operates the social media platform X, formerly known as Twitter. According to X Corp., all users who register for an X account, and/or view the X website or application, agree to binding Terms of Use (Terms) including, in most pertinent part:

1. an express prohibition against “scraping [X’s] Services in any form, for any purpose without [X’s] prior written consent”; and
2. a requirement that those who want to use X data have to do so through tools that X licenses out (“If you want to reproduce, modify, create derivative works, distribute, sell, transfer, publicly display, publicly perform, transmit, or otherwise use the Services or Content on the Services, you must use the interfaces and instructions we provide, except as permitted through the Services, these Terms, or the [developer] terms[.]”).”

Bright Data is a data-scraping company that offers datasets from data that Bright Data scrapes itself, tools that customers can purchase to scrape data themselves, and a service that allow customers to scrape data through proxy servers.

X Corp. brought a number of claims against Bright Data, including breach of contract, tortious interference with contract, and misappropriation. Those claims centered on two activities alleged to violate X’s Terms: (1) Bright Data’s scraping and selling of publicly available content on X, and enabling others to do so; and (2) Bright Data improperly accessing X Corp’s systems and assisting others to do so. Bright data moved to dismiss these claims.

Court Holds That X Corp.’s Data Scraping Claims Are Preempted by Copyright Law

Section 301 of the Copyright Act preempts state law claims that enforce rights that are “equivalent to any of the exclusive rights within the general scope of copyright” and that “come within the subject matter of copyright.” In assessing whether preemption exists, courts have typically looked to whether the state law claim includes an “extra element” that makes it “qualitatively different” from a copyright infringement claim. In recent years, there has been a circuit split as to whether the mere existence of a contractual provision provides that extra element.¹

The Bright Data court took a different, and broader, approach to the question of preemption. Citing the Ninth Circuit’s decision in Ryan v. Editions Ltd. W., Inc.,² the court identified two types of preemption: the “express” preemption set forth in Section 301 of the Copyright Act (i.e., are the state law rights equivalent to any of the exclusive rights within the general scope of copyright) and “conflict” preemption (i.e., does the state law conflict with or undermine the Copyright Act).

Although the court acknowledged that most preemption cases have focused on express preemption, it noted that conflict preemption is the more appropriate basis for analysis when analyzing a party’s rights under a breach of contract claim. The court held that, if it allowed X Corp.’s claims to proceed, X Corp. would effectively be “entrench[ing] its own private copyright system that rivals, even conflicts with, the actual copyright system enacted by Congress,” effectively “exercising a copyright owner’s right to exclude where it has no such right.”

The court identified three ways in which it asserted X Corp.’s state law data scraping claims “undermine the purpose and intended effects of the Copyright Act.” Per the court:

• Since X Corp. is merely a non-exclusive licensee of user-owned X content, allowing X Corp. to prohibit others from using that content would give it greater rights than it is entitled under the Copyright Act. In support of this holding, the court observed that the non-exclusive license granted by users to X Corp. in the Terms mirrored a grant of copyright rights;
• Allowing X Corp. to enforce state law claims would prevent Bright Data and its customers from making a fair use of the copyrighted content on the X platform as is permitted under the Copyright Act; and
• Allowing X Corp. to enforce its state law claims would give X Corp. de facto copyright ownership in content that may not be protected by the Copyright Act (such as short comments), thereby shrinking what would otherwise be in the public domain.

The court acknowledged that in cases where there is a “substantial state law interest” outside of copyright (such as protecting privacy rights), a claim would not be preempted.

Court Holds That X Corp.’s Access Claims Fail To State a Claim

The court also dismissed X Corp.’s various claims relating to Bright Data’s improper access to X Corp. systems, including trespass to chattels, unfair competition and tortious interference with contract. While the court acknowledged that various claims can be triggered by improper access to a third-party system, X Corp. had failed to adequately plead any damage to its servers or business sufficient to sustain these claims.

Of particular note, the court also held that the sale of IP proxies is not “inherently deceptive” since there is no “affirmative duty to identify oneself with a given IP address.” The court noted that use of different IP addresses could, in certain cases, support an unfair competition claim, but suggested that a showing of underlying harmful activity would be required.

Takeaways

• More data scraping defendants likely will argue “conflict preemption.”Undoubtedly, the most important takeaways from the court’s decision are its broad reading of conflict preemption and its statement that unauthorized scraping cases likely are best resolved through the Copyright Act. The decision arguably constitutes a significant departure from many years of jurisprudence that focused on the extent to which a breach of contract claim presented an ”extra element” and was therefore not preempted by the Copyright Act. If other courts follow in Bright Data’s footsteps, and/or the current decision is not reversed on any appeal, it could make it more difficult for companies to rely on breach of contract and state law claims to block unauthorized scraping, where the content at issue is protected (or arguably protected) by copyright. That said, an important factor here was that X Corp. was only a non-exclusive licensee of the applicable content, and it is possible that the court may have reached a different conclusion if X Corp. owned (or held an exclusive license to) the content at issue.
• Terms of use may be revised to eliminate limitations that would conflict with copyright law.As noted, the court highlighted its concern that applying the X platform Terms would have allowed X Corp. to protect content that is not copyrightable, such as short phrases, and would prohibit permissible uses, such as those protected by fair use. Companies may want to consider whether limitations on the use of content set forth in their terms of use make clear that such limits do not prohibit uses otherwise permitted by law.
• Actual knowledge may make browsewrap agreements enforceable.Browsewrap agreements are online terms that are presented through an inconspicuous link, often at the bottom of a web page, and without any requirement that the user manifest assent, such as by clicking an “I agree” button. Courts generally have held that these agreements are unenforceable, as they do not create a binding agreement. However, the Bright Data court noted an important exception to this principle; namely, that browsewrap agreements can be enforceable where the party had actual knowledge of the terms. The court found that this was the case here, and also highlighted that Bright Data was a sophisticated party that was well aware of the contractual prohibitions at issue.

ENDNOTES

1 Compare, for example, ProCD, Inc. v. Zeidenberg, 86 F.3d 1447, 1454–55 (7th Cir. 1996) (“a simple two-party contract is not ‘equivalent to any exclusive rights within the general scope of copyright’ and therefore may be enforced”) and MLGenius Holdings LLC v. Google LLC, No. 20-3113, 2022 WL 710744 (2d Cir. Mar. 10, 2022), cert. denied, 143 S.Ct. 2658 (2023) (holding that certain breach of contract claims may not be qualitatively different from a copyright claim and could therefore be preempted).

2 786 F.3d 754, 760 (9th Cir. 2015).

This post comes to us from Skadden, Arps, Slate, Meagher & Flom LLP. It is based on the firm’s memorandum, “District Court Adopts Broad View of Copyright Preemption in Data Scraping Case,” dated May 21, 2024, and available here.