Law vs. AI (Mostly OpenAI)
Update on the various OpenAI pending Cases
Although OpenAI’s ChatGPT sparked the wider non-tech audience’s interesting AI, many other companies have entered the market with similar tools and LLMs.
Anthropic - Claude
Google DeepMind - Gemini
Mistral AI - Mistral, Mixtral
Meta (Facebook) - Llama
Cohere - Command R
xAI - Grok
Amazon AWS - Titan
IBM - Watson
While the majority of the legal scrutiny has hit OpenAI, these other competitors are watching the outcomes of these cases. The impact OpenAI will be the same for the rest as their creation and operation undoubtedly presents the same concerns for potential litigants. The plaintiffs in the OpenAI cases are alleging a range of violations including misuse of copyrighted material and personal data to train its large language models (LLMs).
These cases represent the frontier in defining the boundaries of fair use, copyright infringement, and data privacy in the AI (especially LLM) era.
This post is attempting a snapshot in time to synthesize the latest developments in each of these lawsuits as of February 21, 2025.
The Copyright Conundrum: A Wave of Infringement Claims
OpenAI’s method of training its LLM sparked litigation predominantly from copyright holders asserting their works were scraped and ingested without consent to train ChatGPT and its predecessors. The core legal issue is whether OpenAI’s ingestion of vast datasets—often publicly available but copyrighted—constitutes fair use under 17 U.S.C. § 107 or direct infringement under § 106. Plaintiffs have argued that the transformative nature of AI outputs does not absolve OpenAI of the violation that is the initial unauthorized reproduction, while OpenAI argues nearly exclusively their use of this training data constitutes fair use. (See Authors Guild v. Google, 804 F.3d 202 (2d Cir. 2015) an example of the analogy that training LLMs is a non-expressive, transformative act.
Tremblay v. OpenAI*: Authors Take the Lead
Filed in June 2023 in the Northern District of California (Tremblay v. OpenAI, Inc.*, No. 3:23-cv-03223-AMO), Tremblay remains a bellwether for author-driven suits. Plaintiffs Paul Tremblay and Mona Awad, representing a proposed author class, allege OpenAI harvested their books to train ChatGPT. Some of their evidence thus far is outputs from ChatGPT that produce accurate book summaries. Recent filings (e.g., Doc. #337, Feb. 20, 2025) reveal ongoing discovery disputes, with plaintiffs seeking detailed insight into OpenAI’s training datasets. Judge Araceli Martínez-Olguín has allowed direct infringement claims to proceed (Order on Motion to Dismiss, July 12, 2024, Doc. #245), while dismissing others (negligence, unjust enrichment). The case hinges on whether OpenAI can substantiate its fair use defense—a question likely destined for appellate review given the paucity of AI-specific precedent. These cases, no matter who prevails at the lower levels, will ultimately land at the U.S. Supreme Court. Tremblay v. OpenAI, Inc., No. 3:23-cv-03223-AMO (N.D. Cal. filed June 28, 2023).
The New York Times v. OpenAI: Media Giants Strike Back
Initiated in December 2023 in the Southern District of New York (The New York Times Co. v. OpenAI, Inc., No. 1:23-cv-11195-SHS), The New York Times v. OpenAI* (with Microsoft as co-defendant) escalated the stakes by pitting a media titan against a technology titan. The Times alleges that OpenAI copied millions of its articles, with ChatGPT reproducing near-verbatim excerpts when prompted (Complaint, Doc. #1, ¶ 2). A November 2024 filing highlighted a dramatic twist: OpenAI inadvertently deleted 150 hours of Times-curated training data evidence due to a “glitch,” though counsel conceded no intent (Maisel Declaration, Nov. 20, 2024, Doc. #189). This mishap complicated discovery and may bolster the Times’ argument that OpenAI’s opacity undermines fair use claims. Judge Sidney H. Stein has yet to rule on OpenAI’s motion to dismiss (Doc. #87, Feb. 26, 2024), but the case’s billion-dollar damages claim underscores its potential to reshape AI liability. (The New York Times Co. v. OpenAI, Inc.*, No. 1:23-cv-11195-SHS (S.D.N.Y. filed Dec. 27, 2023).
The Intercept Media v. OpenAI: A DMCA Twist
A newer front opened in February 2024 (The Intercept Media, Inc. v. OpenAI, Inc., No. 1:24-cv-01515-JR, S.D.N.Y.), where The Intercept accused OpenAI of violating the Digital Millennium Copyright Act (DMCA), 17 U.S.C. § 1202, by stripping copyright management information (CMI) from its articles during training (Complaint, Doc. #1, ¶ 35). Unlike traditional infringement suits, this case sidesteps fair use debates, focusing on statutory harm. On February 20, 2025, Judge Jed Rakoff denied OpenAI’s motion to dismiss the CMI removal claim, finding plausible injury (Order, Doc. #45). This diverges from Judge Colleen McMahon’s November 2024 dismissal of similar claims by Raw Story and AlterNet (Raw Story Media, Inc. v. OpenAI, Inc.*, No. 1:24-cv-01514-CM, S.D.N.Y., Order, Doc. #62, Nov. 7, 2024), signaling a judicial split that could tee up Second Circuit clarification and ultimately, again, a U.S. Supreme Court review. (The Intercept Media, Inc. v. OpenAI, Inc.*, No. 1:24-cv-01515-JR (S.D.N.Y. filed Feb. 28, 2024)).
Beyond Copyright: Privacy and Contractual Challenges
While copyright dominates the headlines, OpenAI also faces litigation over privacy and contractual breaches which expands the liability for them and all the others listed above.
Cousart v. OpenAI: Privacy Claims Falter, Then Pivot
In the Northern District of California (Cousart v. OpenAI, LP, No. 3:23-cv-04557-VC), is a proposed class action alleging that OpenAI and Microsoft violated consumer privacy laws by scraping personal data. The case was dismissed in May 2024 (Order, Doc. #78, May 24, 2024). Judge Vince Chhabria criticized the 204-page complaint’s “distracting allegations,” but granted Plaintiffs leave to amend. Plaintiffs refiled, narrowing their focus to California’s Invasion of Privacy Act (CIPA), Cal. Penal Code § 630 et seq., and unjust enrichment (Amended Complaint, Doc. #92, July 15, 2024), but the case remains a long shot absent concrete evidence of individual harm—a recurring hurdle in AI privacy litigation. Cousart v. OpenAI, LP*, No. 3:23-cv-04557-VC (N.D. Cal. filed Aug. 30, 2023).
Musk v. OpenAI: A Mission-Driven Dispute
Elon Musk’s suit (Musk v. OpenAI, Inc., No. CGC-24-612629, Cal. Super. Ct., San Francisco Cty.), filed in March 2024 and expanded in November 2024 to include Microsoft and others (Amended Complaint, Doc. #35, Nov. 15, 2024), alleges that OpenAI abandoned its nonprofit mission by monetizing ChatGPT, breaching various fiduciary duties. While not a training data case per se, Musk’s claims indirectly challenge OpenAI’s data practices by questioning the ethics of its commercial pivot. OpenAI’s counter-filing cites Musk’s own emails proposing a Tesla merger, framing the suit as “revisionist history” (Answer, Doc. #42, Dec. 10, 2024). The case’s contractual focus offers a rare glimpse into OpenAI’s governance, though its relevance to training data is tangential. Musk v. OpenAI, Inc.*, No. CGC-24-612629 (Cal. Super. Ct., San Francisco Cty., filed Mar. 1, 2024).
Global Dimensions: India’s ANI Suit
OpenAI’s legal woes extend internationally, with ANI v. OpenAI (Delhi High Ct., filed Nov. 2024, Case No. pending) marking India’s first major AI copyright suit. ANI alleges unauthorized use of its news content, seeking data removal and damages (Plaint, ¶ 12). OpenAI’s January 2025 response argues that U.S. preservation obligations (stemming from domestic litigation) preclude compliance, while contesting jurisdiction due to its lack of Indian presence (Written Statement, Jan. 23, 2025, ¶ 18). This clash of legal regimes highlights the extraterritorial challenges AI firms face as global content creators assert rights. ANI v. OpenAI (Delhi High Ct., filed Nov. 2024) (specific case number unavailable as of Feb. 21, 2025).. There is also new uncertainty as to the viability of this case given the recent election and a new administration which has been outspoken multiple times about its interest in ensuring AI dominance in the United States.
President Trump Announces AI Infrastructure Investment. Watch on YouTube
Vice President JD Vance’s Speech at the Artificial Intelligence Action Summit
President Trump’s AI Policy Announcement
Procedural and Strategic Takeaways
Discovery Battles: The Data Veil
Across these cases, discovery remains a flashpoint. OpenAI’s reluctance to disclose training datasets—evidenced by the Tremblay disputes (Tremblay v. OpenAI, Inc., No. 3:23-cv-03223-AMO, Doc. #337) and the NYT data deletion (The New York Times Co. v. OpenAI, Inc., No. 1:23-cv-11195-SHS, Maisel Decl., Doc. #189)—pits transparency against trade secret protections. Courts are grappling with how to balance these interests, with secured inspection protocols (e.g., Silverman v. OpenAI, Inc., No. 3:23-cv-03416-AMO, N.D. Cal., Order, Doc. #156, Oct. 3, 2024) offering a potential model. Lawyers advising AI clients should anticipate heightened judicial scrutiny of data provenance.
Fair Use’s Evolution
The fair use doctrine is under strain. OpenAI’s defense—that training is non-expressive and thus transformative—faces skepticism from plaintiffs who argue commercial outputs (ChatGPT subscriptions) undermine the Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569 (1994) framework. A circuit split seems inevitable, with implications for AI development far beyond OpenAI. Campbell v. Acuff-Rose Music, Inc.*, 510 U.S. 569 (1994).
Settlement Trends
Amid litigation, OpenAI has inked licensing deals with publishers like The Atlantic and Condé Nast, suggesting a dual-track strategy: fight in court, settle out of it. These confidential agreements imply a hybrid model where AI firms pay for data access. However, it is unlikely any one firm could afford to compensate all copyright holders and still produce viable LLMs.
What’s Upcoming for OpenAI’s litigation?
As of February 21, 2025, OpenAI’s training data lawsuits are a legal kaleidoscope—copyright, privacy, and contractual claims refracting through U.S. and global courts. For lawyers, the stakes are twofold: advising clients on AI compliance amid uncertain norms, and anticipating a seismic shift in IP law. The Intercept ruling and others signal that courts are willing to let these cases mature, potentially clarifying fair use’s outer limits. Yet, with appeals looming and international suits like ANI’s adding complexity, a global legal resolution to the various claims facing OpenAI (and be extension all other LLMs) remains uncertain.
Why should I care about these cases?
If courts side with plaintiffs in one or more of these lawsuits, you may need to warn clients about using AI-generated text, images, or code in their work, advertising, website design, graphic design, etc. that could infringe on others’ rights. Businesses are increasingly using AI to draft contracts, legal memos, and compliance documents. If OpenAI (or a similar company) is held liable for errors in their legal output, it could create new professional risks for lawyers relying on AI-powered tools. It could also present additional potential defendants in your lawsuits.
AI is increasingly used in hiring, promotions, and performance evaluations whether you represent a business or employee or rejected candidate understanding the effect of these precedents will inform complaint and answer drafting in related cases.
As I have said to so many others, lawyers and non-lawyers, it is not essential that we all become AI software developers. But, it is important to stay aware of the potential claims and liabilities our clients will encounter and how the courts are more broadly analyzing those issues for a range of cases that many lawyers will confront in their practice.