OpenAI faces copyright backlash as deals mount

OpenAI is embroiled in a series of complex procedural disputes in U.S. courts, notably facing allegations of misleading discovery and evidence concealment in high-profile copyright lawsuits such as the New York Times case. This litigation, which has already cost the Times over $20 million and spans nearly three years, underscores the contentious nature of AI-related intellectual property battles where plaintiffs accuse AI companies of unauthorized use of copyrighted content to train models like ChatGPT. The procedural wrangling reveals the difficulties courts face in managing discovery and enforcing transparency amid rapidly evolving AI technologies.

The legal challenges extend beyond copyright infringement to include trade secret disputes, as exemplified by Apple's lawsuit against OpenAI involving allegations of confidential information leaks by former Apple employees now working at OpenAI. This case highlights the procedural complexities in proving trade secret misappropriation, with Apple struggling to specify which files were accessed despite evidence of unauthorized system access. OpenAI’s unconventional tactic of publicly releasing internal communications to counter Apple's injunction motion further illustrates the heightened tensions and strategic maneuvering characteristic of AI-related litigation.

The surge in AI-related lawsuits in the U.S. reflects a broader legal reckoning with the disruptive impact of AI on intellectual property rights, with copyright, trademark, and image-rights disputes comprising 65.5% of the 131 cases tracked by mid-2025. OpenAI stands at the center of this maelstrom, facing 41 lawsuits including the landmark New York Times case alleging billions in damages. This proliferation of litigation involves a diverse array of plaintiffs—from individuals to government entities—highlighting the multifaceted and unprecedented challenges courts encounter in adjudicating AI’s legal risks and balancing innovation with protection of creative works.

The Indian judiciary has set a landmark precedent by ruling in favor of OpenAI in the ANI Media copyright lawsuit, recognizing AI training as a form of 'fair dealing' under Section 52 of the Copyright Act, 1957. This decision, notably by Justice Amit Bansal of the Delhi High Court, interprets the training of large language models (LLMs) as a transformative research process rather than traditional reproduction, marking a pivotal international stance on AI copyright issues amid an absence of explicit statutory provisions for text and data mining in India.

The ANI Media v. OpenAI case exposes significant gaps and ambiguities within Indian copyright law regarding generative AI, as Section 52 currently lacks clear provisions addressing commercial AI training. This legal uncertainty has prompted interventions from diverse stakeholders—including the Digital News Publishers Association and the Indian Music Industry—highlighting the dispute’s evolution from a bilateral lawsuit into a broader policy debate about intellectual property definitions in the AI era and the potential need for legislative reform.

OpenAI’s defense strategy, emphasizing that LLM training involves statistical token generation rather than direct reproduction of copyrighted works, challenges conventional copyright interpretations not only in India but globally. The Delhi High Court’s extension of Indian copyright law’s reach to extraterritorial conduct—based on substantial links such as ANI’s principal business presence and content access in India—further complicates enforcement dynamics in cross-border AI training disputes.

While the Court distinguished between memorization during AI training and real-time retrieval of copyrighted content, it left unresolved whether live retrieval constitutes 'communication to the public' under Section 2(ff) of the Copyright Act. This nuanced differentiation underscores ongoing legal uncertainties and signals that future rulings will be critical in defining the boundaries of AI-generated content and copyright infringement in India’s evolving jurisprudence.

Anthropic's landmark $1.5 billion settlement over training its Claude AI on 7 million pirated books—covering 500,000 works with 91% author participation—exposes deep ethical contradictions in the AI industry. While companies like Anthropic and OpenAI argue they can train models on any creator’s output if they purchase a single copy, they simultaneously oppose others using their AI-generated content for training, a stance critics label 'completely hypocritical.' This case underscores calls for content providers to unite and collectively negotiate fair compensation to avoid being marginalized as seen in other digital content arenas.

The AI copyright landscape is further complicated by market dynamics and regulatory challenges, as dominant players like Anthropic and OpenAI face accusations of regulatory capture that threaten open source innovation. Meanwhile, surging AI capital expenditures have unsettled stock valuations of tech giants such as Google and Tesla, reflecting investor concerns over escalating development costs and intellectual property risks. These financial pressures intertwine with ethical debates on AI training data usage, content licensing, and the balance between fostering innovation and protecting creators’ rights.

Publishers are grappling with significant referral traffic declines—up to 33% year-over-year—as AI-driven search reshapes content discovery, prompting legal actions against AI firms like OpenAI and Microsoft for unauthorized content use. In response, regulatory bodies such as the UK’s CMA have mandated greater transparency and opt-out controls for publishers, while new business models emerge, including content-for-training marketplaces developed by companies like Parallel AI and Cloudflare. These shifts raise critical questions about pricing, attribution, and publishers’ negotiation power in an evolving ecosystem.

The murky legal status of AI-generated content adds another layer of complexity, as U.S. law denies copyright protection to purely AI-generated works lacking meaningful human input, making them vulnerable to copying and resale. Successful monetization increasingly depends on human creative contributions—such as editing and restructuring—to establish defensible copyrights. Platforms like Amazon KDP have tightened enforcement of AI-content disclosure rules through 2025 and 2026, compelling creators to meticulously document their creative processes to comply with evolving standards and safeguard their intellectual property rights.

Major media companies like News Corp aggressively pursue lawsuits against AI firms refusing to pay licensing fees, while simultaneously striking deals with compliant companies, illustrating a strategic blend of legal pressure and profit motives. The News/Media Alliance’s new $2 billion annual AI licensing agreement—growing 50% year-over-year—extends access to smaller publishers, potentially offsetting traditional revenue declines. However, skepticism remains about whether these licensing revenues will be substantial or merely serve as PR for AI firms historically profiting from uncompensated publisher content, highlighting the fraught ethical and economic tensions at the heart of AI copyright disputes.