Web scraping lawsuits have surged since 2024 as content owners, social media platforms, and news publishers try to stop AI companies and data brokers from harvesting their material, and the live question in nearly every new case is which legal theory still works. Courts have narrowed the Computer Fraud and Abuse Act, rejected a major contract-based claim on copyright preemption grounds, and begun sketching the outer limits of fair use for AI training. Plaintiffs have adjusted: the Digital Millennium Copyright Act’s anti-circumvention provision is now the tool of choice, state-law claims survive when carefully pleaded, and a parallel market in paid crawling access is beginning to form.
DMCA Section 1201 Is Now the Lead Theory
The pivot to the DMCA followed a loss. In May 2024, a federal court in the Northern District of California dismissed X Corp.’s contract-based scraping claims against Bright Data, ruling that enforcing terms-of-service restrictions on scraping would “entrench [X Corp.’s] own private copyright system” in conflict with the Copyright Act.1CNBC. Elon Musk’s X Loses Lawsuit Against Bright Data Over Data Scraping That preemption reasoning pushed plaintiffs to find footing outside contract law, and Section 1201 of the DMCA, which prohibits bypassing technological measures that control access to copyrighted works, has filled the gap.
Reddit has pressed the theory hard. In October 2025, it sued SerpApi, Oxylabs, AWMProxy, and Perplexity AI in the Southern District of New York, alleging the defendants used fake identities, rotating IP addresses, proxy networks, and forged credentials to bypass Reddit’s technical and contractual defenses.2Reuters. Reddit Sues Perplexity, Scraping Data to Train AI System3The New York Times. Reddit Data Scrapers Perplexity Theft Reddit’s chief legal officer described the fight as one against an “industrial-scale ‘data laundering’ economy.”
Google followed in December 2025, suing SerpApi in the Northern District of California and alleging the company used hundreds of millions of fake Google search requests to bypass security protections and scrape copyrighted search results for resale.4Reuters. Google Lawsuit Says Data Scraping Company Uses Fake Searches to Steal Web Content YouTube creators have piled on. In November 2025, Ted Entertainment (operator of the h3h3 Productions channel), Matt Fisher, and Golfholics sued Nvidia in the Northern District of California, alleging the chipmaker bypassed YouTube’s access controls to scrape videos for training its Cosmos AI model.5Piracy Monitor. Nvidia Sued for Allegedly Scraping Copyright Protected Video From YouTube Nvidia moved to dismiss in February 2026, arguing that the DMCA does not prohibit circumventing measures designed to prevent copying, because doing so would undermine fair use.6Law360. Nvidia Says YouTubers AI Scraping Suit Undermines Fair Use Similar creator suits have landed against Snap, Meta, and ByteDance (over its MagicVideo model) in early 2026.7Copyright Alliance. AI Copyright Lawsuit Developments
Whether this entire line holds up may turn on a pending Second Circuit appeal. In Yout LLC v. Recording Industry Association of America, the court heard oral argument in February 2024 on the scope of Section 1201’s access-control provisions. As of mid-2026 it has not ruled, though it accepted an amicus brief from Suno and Udio in October 2025 and allowed supplemental briefing into early 2026.8CourtListener. Yout LLC v. Recording Industry Association of America Inc A decision would be the first appellate guidance on the anti-circumvention framework now underlying most new scraping suits.
State-Law Claims Can Survive Copyright Preemption
Reddit’s separate suit against Anthropic, filed in San Francisco Superior Court in June 2025, tested whether a platform without copyright in its users’ posts can still sue scrapers. Reddit alleged Anthropic trained its AI models on Reddit data while bypassing robots.txt directives and IP rate limits, bringing only state-law claims: breach of contract, unjust enrichment, trespass to chattels, tortious interference, and unfair competition.9Courthouse News Service. Reddit v. Anthropic Remand Order
Anthropic removed the case to federal court and raised the same preemption defense that worked for Bright Data. On March 28, 2026, Judge Trina L. Thompson rejected it and sent the case back to state court, finding Reddit’s claims contained “extra elements” that are “qualitatively different” from copyright infringement: contractual obligations under Reddit’s User Agreement, the bypassing of technical safeguards, and the protection of user privacy rights.9Courthouse News Service. Reddit v. Anthropic Remand Order The practical takeaway for platforms: strong terms of service combined with real technical protections may support state-law claims that would have failed on a bare contract theory.
Fair Use and AI Training
No appellate court has yet decided whether scraping copyrighted material to train AI models is fair use. Two 2025 district court rulings have begun to mark the edges.
In Kadrey v. Meta Platforms, Inc., Judge Vince Chhabria of the Northern District of California granted summary judgment for Meta in June 2025, calling the company’s use of copyrighted books to train its Llama language models “highly transformative.” The judge was blunt about what the ruling did not do, writing that the plaintiffs had “made the wrong arguments and failed to develop a record in support of the right one.”10Skadden. Fair Use and AI Training The decision binds only the thirteen named plaintiffs. A separate claim that Meta distributed pirated copies of books through BitTorrent “seeding” remains active.11Justia. Kadrey et al v. Meta Platforms Inc
In Bartz v. Anthropic, a court likewise found model training “exceedingly transformative” in June 2025, but ruled that Anthropic’s downloading of books from pirate sites was itself infringing. The case settled in September 2025 for approximately $1.5 billion, with Anthropic paying roughly $3,000 for each of 482,460 books it had downloaded from pirate libraries.7Copyright Alliance. AI Copyright Lawsuit Developments The combined signal from these two cases: how the training data was obtained can matter as much as what the model does with it.
A May 2025 U.S. Copyright Office report, “Copyright and Artificial Intelligence, Part 3: Generative AI Training,” pushed back on broad transformativeness claims. The Office argued that if a model is trained to produce content aimed at the same audience as the originals, the use is “at best, modestly transformative,” flagged “market dilution” as a cognizable harm even when outputs are not identical to any training input, and said Retrieval-Augmented Generation technology is “very likely infringing.”12Copyright Alliance. Copyright Office’s AI Report Takeaways
The first appellate test is close. The Third Circuit heard oral argument on June 11, 2026, in Thomson Reuters v. Ross Intelligence, an appeal from a district court ruling that Ross Intelligence’s use of Westlaw headnotes to train an AI legal research tool was not fair use.13CourtListener. Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc Observers expect it to be the first appellate ruling on fair use applied to a commercial AI product.
The most watched trial-court case is The New York Times Co. v. Microsoft Corp., filed in December 2023 in the Southern District of New York. The Times alleges OpenAI and Microsoft used millions of its articles without permission to train GPT models and that ChatGPT can reproduce near-verbatim passages, letting users bypass paywalls.14AI Lawsuit Tracker. New York Times v. OpenAI In March 2025, Judge Sidney Stein allowed direct and contributory copyright infringement claims to proceed while dismissing certain DMCA Section 1202 and common-law unfair competition claims.15Reuters. Judge Explains Order in New York Times-OpenAI Copyright Case Summary judgment briefing concluded in April 2026, with a ruling expected in the third quarter of 2026 and any trial likely in late 2026 or 2027.
The broader docket keeps growing. More than 70 copyright suits against AI companies were ongoing as of mid-2026.7Copyright Alliance. AI Copyright Lawsuit Developments16CourtListener. Disney Enterprises Inc v. Midjourney Inc17Encyclopædia Britannica. Britannica Files Copyright and Trademark Infringement Lawsuit Against Perplexity18Sussman Godfrey. Britannica v. Perplexity Complaint In music, Universal Music Group and Warner Music Group settled with Udio in late 2025, and Warner also settled with Suno, with licensing deals tied to authorized AI music services launching in 2026; Sony continues to litigate against Udio.
Platforms, Terms of Service, and the Limits of Control
Platforms are fighting on two fronts: suing scrapers who hit their own sites, and defending against claims that they scraped others. In January 2024, Judge Edward Chen of the Northern District of California handed Meta a loss in Meta Platforms Inc. v. Bright Data, granting summary judgment for the scraping company. The court held that Meta’s terms of service could not prohibit “logged-off” scraping of publicly available data, because someone scraping without logging in is not a “user” who agreed to those terms. Chen noted that Meta had previously removed a clause stating that merely “accessing” the platform bound someone to its terms, and warned that letting companies control who can collect publicly available data “risks the possible creation of information monopolies.”19Courthouse News Service. Federal Judge Rules Against Meta in Data Scraping Case
LinkedIn has gone the other direction. After the long-running hiQ Labs litigation ended in a confidential settlement, LinkedIn sued ProAPIs in the Northern District of California in October 2025, alleging the company created millions of fake accounts to bypass password protections and scrape user data, then charged customers up to $15,000 per month for access.20The Record. LinkedIn Sues Data Scraping Company It resolved a separate suit against Proxycurl in July 2025.21Bloomberg Law. LinkedIn Battles Online Scrapers in Perpetual Struggle Over Data The distinction that matters: scraping behind a password wall still gives platforms strong claims, while scraping publicly viewable pages does not.
Why the CFAA Rarely Works Anymore
The Computer Fraud and Abuse Act was once the main scraping weapon, and two rulings have largely dismantled that role. In Van Buren v. United States (2021), the Supreme Court adopted a “gates-up-or-down” test, holding that the CFAA reaches only conduct that bypasses a technological barrier, not misuse of data someone is already authorized to access.22Reporters Committee for Freedom of the Press. Scraping Not Violation of CFAA On remand in hiQ Labs v. LinkedIn (2022), the Ninth Circuit applied that framework and concluded that publicly accessible websites have no “gates to lift or lower,” meaning scraping public data likely falls outside the CFAA even when a cease-and-desist letter or terms of service forbids it.23Ninth Circuit Court of Appeals. hiQ Labs Inc v. LinkedIn Corp
The CFAA still has a role when a scraper bypasses password walls or uses stolen credentials, as LinkedIn alleges in ProAPIs. For data any visitor could see in a browser, it is not a viable tool.
Privacy Law and International Enforcement
Scraping that implicates personal data has drawn separate enforcement under privacy law. Clearview AI’s practice of scraping billions of facial images has triggered actions in several jurisdictions. A U.S. class action settlement approved in March 2025 gave class members a 23% equity stake in Clearview, valued at approximately $51.75 million, in a case brought under the Illinois Biometric Information Privacy Act.24Loevy & Loevy. Judge OKs Innovative $51.75 Million Settlement in Clearview AI Class Action Lawsuit The Dutch Data Protection Authority fined Clearview €30.5 million in September 2024 under the GDPR for maintaining an illegal biometric database.25Silicon Republic. Clearview AI Fine Dutch Images Facial Recognition A UK Information Commissioner’s Office fine of £7.5 million was overturned in 2023 after a tribunal found the ICO lacked jurisdiction because Clearview served only foreign law enforcement agencies.26BBC. Clearview AI Overturns UK Privacy Fine
In August 2023, twelve international data protection authorities issued a joint statement affirming that personal data on public websites remains subject to privacy law and recommending rate limiting, bot detection, and other technical defenses against scraping.27Hunton Andrews Kurth. Joint Statement Published on Data Scraping and the Protection of Privacy The Irish Data Protection Commission had earlier fined Meta €265 million for a scraping breach that exposed data belonging to approximately 533 million Facebook users.28TechCrunch. Digital Rights Ireland GDPR Lawsuit Facebook Data Scraping Breach
A Paid-Access Market Starts to Form
While the lawsuits continue, a transactional alternative is taking shape. Cloudflare launched a “Pay per crawl” program in July 2025 letting website owners set a price per page request from AI crawlers. The system uses the HTTP 402 “Payment Required” status code: a crawler receives a price in the response header and can pay or move on, with Cloudflare handling authentication and distributing payments to publishers.29Cloudflare. Introducing Pay Per Crawl As of mid-2026, the program is in closed beta.30Cloudflare. What Is Pay Per Crawl Private licensing deals have moved in parallel, with Google and OpenAI entering agreements with platforms including Reddit, suggesting the market is shifting toward paid access whether companies arrive there by choice or by litigation pressure.