Fair Use Or Fair Game?

The artificial intelligence gold rush has a new battleground — and it isn’t about chips, models, or valuations.

It’s about books. Movies. Music. News articles.

And whether tech companies were allowed to copy them in the first place.

After a surge of lawsuits and a record-setting settlement in 2025, 2026 is shaping up to be the year U.S. courts decide how copyright law applies to generative AI.

At stake: billions of dollars, the future economics of AI, and who ultimately controls the raw material that powers today’s most advanced systems.

The Trillion-Dollar Question: What Counts As Fair Use?

The core legal issue is deceptively simple: Can AI companies train their models on copyrighted content without permission under the doctrine of fair use?

Companies like OpenAI, Google, and Meta argue that training AI is transformative.

Models don’t store books or movies in traditional ways, they say. Instead, they convert content into statistical relationships that generate new outputs.

However, copyright holders see it differently.

They argue that AI systems are built on massive unauthorized copying — and that the outputs can compete directly with the very creators whose work trained the models.

If courts agree with tech companies, AI development continues largely unchanged. If they side with creators, the industry may need to pay for licenses — potentially costing billions and reshaping profit margins across the sector.

Judges Split: Transformative Tool Or Market Threat?

In 2025, federal judges began issuing early fair use rulings — and the results were mixed.

In one high-profile case, a federal judge ruled that Anthropic’s use of books to train its AI system was quintessentially transformative, leaning in favor of fair use.

But the same judge found the company liable for storing millions of pirated books outside the training process, exposing it to enormous potential damages before a settlement was reached.

In another case, a judge ruled in favor of Meta over its Llama model training but warned that generative AI could, in some circumstances, flood creative markets and undermine incentives for human creators.

Translation: AI companies are winning some battles — but the war is far from settled.

Newsrooms, Hollywood And The Music Industry Push Back

The lawsuits now stretch across nearly every creative sector.

The New York Times sued Perplexity AI, alleging the startup scraped and reproduced paywalled articles through its retrieval-augmented generation system, diverting traffic and revenue.

Meanwhile, Disney, Universal Pictures, and Warner Bros. are battling Midjourney, claiming the AI image generator trained on — and reproduces — iconic characters without authorization.

Major music publishers have also sued Anthropic, alleging pirated compositions were used to train its Claude models.

These aren’t fringe complaints. They strike at the heart of how generative AI systems were built.

The Billion-Dollar Twist: Settlements And Licensing Deals

Not every copyright holder is fighting.

Some are negotiating.

Disney agreed to invest $1 billion in OpenAI and allow character use in its Sora video generator. Music companies have struck partnerships with AI music startups. Thomson Reuters licensed its content to Meta.

This signals a possible middle path: instead of banning AI training, the industry may move toward structured licensing deals — creating a new revenue stream for media companies while raising costs for AI developers.

For investors, that matters.

Higher licensing costs could squeeze margins. But clear legal guardrails could also reduce uncertainty — something markets tend to reward.

Why 2026 Could Be Decisive

Multiple hearings and rulings are expected this year involving Anthropic, Google, Stability AI, Suno, and others.

The outcomes could cement broad fair use protections for AI training, force companies into expensive licensing regimes, or create a patchwork of rulings that prolong uncertainty.
Either way, the legal clarity coming — or not coming — in 2026 could define the next phase of the AI boom.

Because this isn’t just about whether a chatbot can summarize an article.

It’s about who owns the building blocks of artificial intelligence — and whether the machines were trained fairly.

Or whether they’ve been playing a fair game all along.

That's all for this week! If you found these updates useful, you'll like more from this newsletter. Get deeper dives, hot takes, and all the latest tech news delivered straight to your inbox.