News outlets demand new copyright rules for AI training

AI models have been consuming the work of journalists around the clock to get smarter, often without paying a dime or giving any credit. Now, the media industry is drawing a collective line in the sand.

A powerhouse coalition of global news outlets and publishers published an open letter calling on lawmakers worldwide to rein in generative AI developers and protect copyrighted training data.

Demanding consent, cash, and transparency

The list of signatories reads like a who's who of traditional media: The Associated Press, Agence France-Presse, Getty Images, Gannett, the European Publishers’ Council, and The Authors Guild, among others.

They are pushing global regulators for several strict mandates. At the top of the list: full transparency into training datasets, mandatory consent from copyright holders before any data is ingested, and the legal right to negotiate licensing fees with AI operators.

They also want governments to require clear labeling for AI-generated material and legally mandate that AI platforms eliminate bias and misinformation from their systems.

According to the letter, foundation models ingest media reporting and redistribute information without attribution, remuneration, or basic consideration. The publishers argue this systematically dismantles the core business models of media—namely subscriptions, licensing, and advertising—which in turn jeopardizes public access to reliable news.

Fear, lawsuits, and secret deals

The timing is hardly accidental. Google recently pitched its AI news-writing assistant, Genesis, to executives at The New York Times, The Washington Post, and News Corp. Meanwhile, early publisher experiments with AI-written articles have resulted in high-profile, embarrassing factual errors.

Publishers are far from alone in this battle. The Senate has held multiple hearings on AI copyright, visual artists are suing Midjourney and Stable Diffusion, and comedian Sarah Silverman filed a copyright infringement suit against OpenAI.

Yet, the media's stance is more nuanced than outright resistance. The letter notes generative AI offers significant potential benefits, and publishers simply want a seat at the table.

In fact, Reuters pointed out that some signatories are already playing both sides. The Associated Press had recently struck a licensing deal with OpenAI, trading access to its historical news archive in exchange for tech access and compensation.

Why it matters

This isn't an existential boycott against AI—it's the opening salvo of a massive commercial negotiation.

For publishers and independent creators, training transparency is the holy grail. Without legally mandated dataset audits, creators have virtually no leverage to prove their work was stolen, let alone demand a cut. Collective bargaining is their only real shot against Big Tech's bottomless wallets.

For the AI industry, the era of scraping the open web with impunity is quickly coming to an end. If lawmakers enforce mandatory consent, training cutting-edge foundation models will become significantly more expensive. While tech giants can afford to write eight-figure licensing checks, open-source developers and cash-strapped AI startups could easily get priced out.

We think the media's collective front also reveals an obvious crack. When major players like the AP sign angry letters while quietly pocketing licensing deals from OpenAI, it shows Big Tech will likely divide and conquer. The biggest newsrooms will get paid, while smaller outlets and freelance creators risk being left behind.