Major copyright lawsuits against OpenAI and Microsoft allege that large language models were trained on millions of copyrighted works without permission or licensing. If your company uses ChatGPT, Copilot, or similar AI tools, you may have inherited some of this legal uncertainty as a downstream user, even if you weren't directly involved in the dispute. Understanding the real shape of this litigation, and what it means for your AI vendor contracts, matters now.

What Are Companies Actually Being Sued Over?

The core allegation is straightforward: large language models (LLMs) are trained on enormous datasets scraped from the internet, and a significant portion of that material consists of news articles, books, and other copyrighted works used without a license or explicit permission from rights holders. Plaintiffs argue this constitutes copyright infringement at a scale the legal system has never processed before.

Pulitzer Prize-winning authors, including journalists who covered major historical events, have joined lawsuits alleging what their court filings call "rampant theft" of copyrighted material. The New York Times has filed its own litigation against Microsoft and OpenAI over unlicensed use of news articles to train models. Separately, U.S. Senator Josh Hawley has publicly accused AI developers of what he called the largest intellectual property theft in the country's history, in comments tied to congressional scrutiny of AI training practices.

Defendants, primarily OpenAI and Microsoft in the highest-profile cases, generally argue this falls under "fair use," a legal doctrine that permits certain unlicensed uses of copyrighted material, particularly for purposes like research, commentary, or transformation into something sufficiently new. None of the active litigation has reached a final resolution that definitively settles whether AI training on copyrighted material constitutes infringement or falls under fair use, and that uncertainty is likely to persist for at least another year or two given the pace of U.S. court proceedings.

How Should Businesses Protect Themselves Right Now?

• Check for IP Indemnification: Some enterprise AI vendors, including Microsoft in certain enterprise Copilot agreements, have begun offering indemnification clauses that shift copyright liability away from the customer. However, not all vendors or all pricing tiers include this protection, so it's worth checking explicitly rather than assuming it's standard.
• Create an Internal AI Tool Inventory: Many companies don't have a clear internal record of where employees have adopted AI tools informally across departments. Building this inventory makes it easier to assess actual legal exposure if litigation outcomes shift the ground under a specific vendor.
• Treat AI Vendor Selection as a Compliance Decision: Organizations best positioned through this period of legal uncertainty are those that built AI vendor risk into their existing compliance and audit processes, rather than treating AI tools as routine software purchases outside that oversight.
• Review Existing Vendor Contracts: If you've already adopted an AI tool, review your contract to understand what liability sits with you versus the vendor, and whether the agreement addresses IP or data liability explicitly.

Why Does This Matter Beyond the Courtroom?

If you've integrated an AI tool into your operations, you've inherited some of this legal uncertainty as a downstream user, even if you weren't a party to any lawsuit. This raises concrete questions about your company's exposure. If the underlying legal questions get resolved against OpenAI and Microsoft, it could affect licensing costs, tool availability, and legal exposure for every business built on top of them.

The practical lesson for businesses isn't to avoid AI tools, but to bring the same compliance discipline to AI adoption that you would apply to any enterprise software relationship. Companies running Microsoft's enterprise resource planning tools, Microsoft 365 applications, or enterprise workflow automation software as their operational backbone are facing a smaller-scale version of the same question: understanding exactly what you're licensed to do with a vendor's platform, and what liability sits with you versus the vendor, is core to any enterprise software relationship, AI-specific or not.

What Role Does Labor Play in This Dispute?

Alongside the copyright fight runs a parallel, related concern about labor. Content moderation and data-labeling workers, often contracted in lower-wage countries, have described difficult working conditions filtering the toxic content used to train these models. Creative professionals, including writers, illustrators, and translators, have raised concerns that AI systems trained partly on their work now compete with them commercially, without compensation flowing back to them.

Labor unions have engaged directly with these companies. Microsoft entered a formal partnership with the AFL-CIO (American Federation of Labor and Congress of Industrial Organizations) specifically to address worker concerns about AI's impact on jobs, including AI education for union members and structured feedback channels between labor leaders and the company. This represents a negotiated framework for worker input on AI deployment, rather than litigation, but it reflects the broader concern that AI development has created winners and losers in the creative and data-labeling workforce.

What's the Broader Context for U.S. IP Protection?

The AI copyright disputes are unfolding against a backdrop of growing recognition that intellectual property has become the dominant asset for both companies and nations. According to remarks by John A. Squires, Under Secretary of Commerce for Intellectual Property and Director of the U.S. Patent and Trademark Office (USPTO), intangible assets account for approximately 92 percent of the total market value of S&P 500 companies, leaving only 8 percent tied to physical or tangible assets.

IP-intensive industries accounted for 44 percent of U.S. GDP, or $11.4 trillion, in 2024. That amount alone is almost equal to the combined GDPs of Germany, Japan, and France. These industries directly employ almost 50 million Americans, or 33 percent of total U.S. employment. Annual output per worker in IP-intensive industries was nearly $230,000, a 60 percent higher rate of labor productivity than workers in non-IP-intensive industries at $145,000.

The USPTO received 673,000 new trademark applications through the third quarter of 2026, an increase of 11 percent over the same period in 2025, reflecting growing demand for brand protection in the market. Patent examiners process more than 600,000 patent applications every year, and the office is on track to issue its 13 millionth patent in the coming year.

In this context, the AI copyright litigation represents a fundamental question about how intellectual property law adapts to new technologies that can process and learn from vast amounts of copyrighted material at scale. The outcome of these lawsuits will likely shape not just AI development, but how companies across all industries think about data, licensing, and vendor risk for years to come.