For decades, the media industry focused on distribution. Then monetisation took centre stage. Today, a more fundamental question has emerged: who owns the raw material that trains the machines shaping culture, commerce, and information?

Generative AI did not appear from nowhere. It learned from articles, photographs, videos, books, archives, and years of user-generated content. Much of that content belongs to media organisations. As a result, one question now dominates the debate: who gave permission?

What began as quiet concern has grown into lawsuits, licensing talks, and regulatory pressure. Across publishing, journalism, film, and photography, media companies have realised that AI systems have absorbed their work without consent, credit, or payment. The industry is not trying to block AI. It is trying to define the rules that govern how AI learns.

At stake are copyright law, creative ownership, and the long-term economics of content creation.

Generative AI relies on vast datasets. The better the quality, diversity, and credibility of the data, the stronger the system becomes. Newsrooms and publishers spent decades building trusted archives, original reporting, and cultural records. Today, they compete with AI outputs trained, in part, on that same work.

This contradiction has forced a reckoning.

In recent months, leading publishers have taken different paths. The New York Times and Getty Images have filed lawsuits, arguing that AI firms copied and exploited their content at scale. Axel Springer, the Associated Press, and Reuters have chosen licensing deals instead. These agreements allow controlled access to archives in exchange for payment and usage limits.

Both approaches rest on the same belief. Training data has value, and the media industry can no longer treat it as free.

This moment differs from earlier digital disruptions. When search engines and social platforms rose, many media companies reacted too late. They surrendered control over distribution and pricing in exchange for reach. With AI, many have decided not to repeat that mistake.

Media leaders argue that training data is core infrastructure. Control over it means influence over the future value chain.

Regulators are paying attention.

In Europe, policymakers are debating how copyright law applies to machine learning, especially around text and data mining exemptions. In the United States, courts are weighing whether AI training counts as fair use or mass infringement. Governments are also exploring rules that would force AI developers to disclose their training data, a move that could reshape the economics of AI development.

The media industry has become a key voice in these discussions, not because it fears technology, but because it understands leverage. Without high-quality, human-created content, generative AI risks becoming repetitive, inaccurate, and detached from reality. That dependence gives media companies bargaining power, if they choose to use it.

Still, the industry remains divided.

As AI-generated summaries, images, and videos compete for attention, traffic, and advertising spend, the line between original creation and automated reproduction keeps blurring. In journalism, the risk goes beyond revenue. If AI systems trained on past reporting replace original journalism, incentives to fund investigations weaken. Public accountability suffers as a result.

Many media executives say their push for regulation is not about protectionism. It is about sustainability. They want to protect the systems that will shape the future information ecosystem without destroying the institutions that built it.

Reputation also matters. Media brands depend on trust, verification, and accountability. When AI models repeat errors, bias, or misinformation drawn from uncurated data, audiences often blame the original sources. Publishers argue that stronger control over training data helps protect editorial integrity in an age of synthetic content.

This conflict will not end quickly. Lawsuits will drag on. Regulations will face debate, revision, and resistance. Commercial partnerships will expand and then reset. Still, the direction is clear. The era of unrestricted AI training on media content is coming to an end.

Generative AI will keep advancing. The rules around it will evolve too.

For the media industry, the goal is not to stop progress. It is to redefine engagement. To move from passive data suppliers to active participants in the AI economy. If machines are to learn faster, they must learn within boundaries shaped by consent, compensation, and accountability.

AI’s ability to transform media is no longer in doubt. The real question now is whether the media industry will have a say in how that learning happens.

ALSO WATCH MARKETING EDGE ONTV