Two More Newsrooms Sue OpenAI and Microsoft—Here's Why It Matters for Developers

The Lawsuit: What Just Happened

<cite index="1-3">The Seattle Times and Newsday filed suit Friday in federal court, accusing OpenAI of using their published work to train large language models</cite>. <cite index="2-5">The suit alleges that OpenAI and Microsoft scraped the newspapers' websites, including content behind paywalls, and incorporated articles into datasets used to train ChatGPT, Microsoft Copilot and Bing's AI features</cite>.

This isn't a first-time complaint. <cite index="4-5">The New York Times sued OpenAI and Microsoft in 2023 over similar allegations, while other copyright holders have brought cases against AI companies including Anthropic and Meta Platforms</cite>. What's changed is momentum—the lawsuits keep piling up.

The Money Problem Behind the Lawsuit

<cite index="1-5">Attorneys representing the outlets cited industry data that showed search referral traffic to midsize publishers declined by 47% year over year</cite>. That's not coincidence. <cite index="7-3">The newspapers said the companies' AI products can reproduce passages from their reporting, closely paraphrase articles and provide answers that reduce the need to visit their websites or buy subscriptions</cite>.

Consider what happens when someone asks ChatGPT about a breaking news story instead of clicking through to The Seattle Times' article. <cite index="15-9">Training large language models requires vast datasets, yet publishers receive no compensation when their content fuels AI systems that compete with their own business</cite>.

The Irony That Makes This Case Unique

<cite index="5-7">Microsoft and OpenAI have funded some of the Seattle Times' journalism projects and fellowships</cite>. A newspaper suing companies that literally helped pay for its journalism operations sends a signal: the business model conflict has become too serious to overlook, no matter the relationship.

What the Companies Say

<cite index="3-3">OpenAI said its models are trained on publicly available data and grounded in fair use, which helps hundreds of millions of people improve their daily lives</cite>. Fair use is their legal shield—the argument that training LLMs on existing content counts as transformative.

<cite index="12-4">Microsoft argues its Copilot AI rarely reproduces copyrighted content, citing an analysis of 8.2 million conversations where only 24 responses contained matching passages</cite>. But the lawsuit disputes this math, arguing even occasional reproduction combined with commercial use defeats fair use protections.

The Bigger Picture: Courts Are Shifting

<cite index="24-1,24-3,24-4">By early 2026, the era of "train first, ask later" is over. Between 2023 and 2024, over 50 copyright lawsuits were filed against AI companies</cite>. Courts are taking these seriously.

<cite index="23-8,23-9">Thomson Reuters v. Ross Intelligence found that Ross Intelligence's use of Westlaw headnotes to train its legal AI was not fair use, focusing on market harm—Ross's AI competed directly with Westlaw's research services. The ruling established that market competition between an AI's outputs and its training data source is the key inquiry</cite>.

That precedent matters here. News organizations argue that AI summaries don't just coincidentally replace articles—they're explicitly designed to do so.

What Developers Need to Know Right Now

If you're building AI applications, this trend has real implications. <cite index="24-13,24-14,24-15">The US Copyright Office stated that when AI training competes with existing licensing opportunities for original works, the fair use analysis tilts against the AI company. If a chatbot can summarize a news article, that competes with the newspaper's subscription model. Fair use is not a blank check for commercial AI training</cite>.

<cite index="24-17,24-20,24-21">The clearest trend across all cases is the shift from opt-out to opt-in. The legal and regulatory momentum is moving in the opposite direction. UMG v. Udio established opt-in for music</cite>. Books are heading that way too. Assume news content will follow.

If you're training custom models or fine-tuning existing ones, the safest path is licensing data explicitly. Many publishers now offer licensing agreements. The cost of a license is often far cheaper than litigation or forced model retraining.

For developers using mainstream AI APIs like ChatGPT or Copilot in production applications, these lawsuits don't directly affect your code—but they do affect the legal landscape around the tools you depend on. Outcomes could influence pricing, feature availability, or terms of service.

What Comes Next

<cite index="8-8">The newspapers are seeking an unspecified amount of damages, as well as court orders requiring the "impoundment and/or destruction" of copies of their works, training datasets, or AI models that incorporate them</cite>. That's a dramatic ask—not just compensation, but destruction of entire systems. It signals how seriously publishers view existential threats to their business.

These cases won't be resolved quickly. The original New York Times suit is still ongoing. But the pattern is clear: the days of treating the internet as free training data are ending.

If you're considering using external data for AI training or fine-tuning, start building a licensing audit into your workflow today. <cite index="24-8">Landmark settlements and rulings have redrawn the lines around what AI companies can and can't do with other people's work</cite>. Your competitors who haven't acknowledged this shift yet are building legal liability into their products.

You can test how today's AI tools handle sensitive content using NeonCodex AI's sandbox environment, which lets you run queries against different model versions and see exactly what they surface—helpful for understanding competitive output and planning your own approach.

Source: [TechCrunch](https://techcrunch.com/2026/09/05/seattle-times-and-newsday-are-the-latest-publications-to-sue-openai-and-microsoft/)