The UK government is currently navigating one of the thorniest legal issues of the AI era: whether artificial intelligence developers should be allowed to use copyrighted material—books, songs, images, journalism, and more—to train their models, and if so, under what conditions.

A consultation launched in 2024 by the UK Intellectual Property Office (IPO) sought input from industry, creatives, and the public to help shape this policy. Now, in early 2025, the government is moving towards legislative proposals that aim to strike a balance between encouraging AI innovation and protecting the rights of content creators.

But can such a balance actually be struck? That’s the heart of the debate. And as you’ll see, it’s as much about economics and global competitiveness as it is about copyright law.

What the government is proposing

The consultation outlined four broad options for handling the copyright implications of AI training:

  1. Do nothing – Leave the current framework in place, which is vague and open to interpretation.
  2. Require licensing – Make it mandatory for AI companies to obtain permission from rights holders in all cases.
  3. Broad exception – Allow AI companies to mine data freely without needing permission, similar to regimes in Japan and Singapore.
  4. Opt-out system (Option 3) – Allow AI companies to use copyrighted materials unless creators actively opt out, for example via metadata or web protocols like robots.txt.

The government appears to be leaning towards Option 3—the opt-out model—with added transparency and safeguards. This approach would keep the UK competitive while giving creators some control over how their work is used in the development of AI systems.

How the tech giants responded

Two major players—OpenAI and Google—have responded formally to the UK consultation, each issuing detailed statements supporting a pro-innovation stance.

OpenAI argues that text and data mining (TDM) is essential to training large models and should be protected under copyright law. They support a broad exception with opt-out rights and suggest that such a system would benefit both developers and society at large. They also caution against creating confusing or fragmented opt-out mechanisms, warning that poor regulation could stifle innovation.

Google goes even further, pushing for what it calls an “amended Option 3”. Their key points include:

  • AI training is a non-expressive use of content—meaning it doesn’t directly copy or reproduce original works.
  • A broad TDM exception should apply to both non-commercial and commercial AI training.
  • Rights holders should have the ability to opt out, but not the right to demand payment automatically.
  • Excessive transparency rules (like disclosing datasets or training data sources) could expose trade secrets and harm UK competitiveness.

Google is especially keen on framing the issue as an opportunity: get this right, they say, and the UK could become an “AI maker, not an AI taker.”

The backlash from creators

Not everyone is on board. Writers, musicians, designers and other creatives have loudly protested the proposed legislation. High-profile figures like Sir Paul McCartney and Kate Bush have spoken out against what they see as the unlicensed and unpaid use of their intellectual property.

Their concern is straightforward: AI companies are building powerful systems using other people’s work—often created over a lifetime—without asking or paying. Many fear that this will hollow out the creative industries and erode the ability of artists to make a living.

There’s also unease around the idea that opting out requires technical know-how, or that creators must opt out rather than being asked for permission.

What’s likely to happen next

Barring a major political U-turn, the UK government is likely to move forward with a version of Option 3: a framework that allows AI developers to mine publicly available data by default, with tools for rights holders to opt out.

Here’s what to expect:

  • A stronger emphasis on machine-readable opt-out tools, such as robots.txt, which web publishers can use to block AI crawlers.
  • Light-touch transparency requirements to reassure the public without exposing corporate trade secrets.
  • Continued calls for voluntary licensing models, especially in industries like music and news, where rights holders have more bargaining power.

In short, the UK is trying to have it both ways: to keep AI firms happy while offering creators a way to shield their content—if they know how.

The big picture

What’s unfolding here isn’t just a legal technicality—it’s a foundational choice about how we value creativity in the AI age.

For the tech industry, data is like oxygen. Limiting access to it is seen as a threat to progress. But for creators, their work is their livelihood, and many feel that AI is being built on an unacknowledged foundation of their efforts.

The UK’s approach will likely become a test case for other countries watching closely. If it works, the UK may well become a hub for responsible AI innovation. If it fails, it could alienate the creative communities that form a vital part of the country’s cultural and economic life.

We’re at the beginning of this story, not the end—and whatever the government decides, the push and pull between innovation and rights will only intensify in the years ahead.