""As Ebenezer Scrooge can testify, Christmas is a time for reflecting on the past, the present and the future," says Matt Roberson, director of global public policy and platform strategy at the Financial Times. "Like Scrooge, many of the biggest AI developers go into the holiday season with no qualms about extracting commercial value from human labor, in the form of human-authored copyright material, without paying for it. That position is unsustainable and has to change.""
"Naturally, the toothpaste is out of the tube when it comes to AI companies training on publishers' archives, but now the industry is fighting to shape what comes next. Publishers want a fair, structured, regulated AI environment that involves AI companies not just splurging on chips and computing power, but actually coughing up for the information their large language models (LLMs) churn out."
"There have been flickers of hope throughout 2025. Cloudflare helped publishers give the middle finger to unwanted bot blocking - a much-needed slither of leverage to at least stem the flow of unwanted scraping. Amazon entered the AI licensing fray, showing it was willing to pay The New York Times, Condé Nast and Hearst good money in exchange for access to their content. Microsoft's move to establish an AI content marketplace with a group of publishers, showed the media business world that it was willing to pay"
Publishers face widespread unscrupulous AI scraping that undermines their work and revenue. The industry seeks a regulated AI environment requiring companies to pay for content and respect copyright. Some progress emerged in 2025: Cloudflare tools reduced unwanted bot scraping; Amazon and Microsoft struck licensing deals or marketplaces paying publishers; Meta secured multi-year AI licensing deals with multiple major publishers. Industry voices say extracting value from human-authored material without payment is unsustainable. Publishers demand that AI companies contribute beyond hardware investment, paying for information used to generate model outputs.
Read at Digiday
Unable to calculate read time
Collection
[
|
...
]