Future legislation compensating news outlets for content must account for AI: report

A man uses a computer keyboard in Toronto, Monday, Oct. 9, 2023 in this photo illustration. THE CANADIAN PRESS/Graeme Roy

TORONTO — Canadian researchers say any future legislation meant to level the playing field between Big Tech and journalism organizations must account for generative artificial intelligence.

The experts behind a report released this week to analyze Canada’s Online News Act say it’s important for the popular technology to figure into future policies because the news industry’s battleground for fairness and compensation is shifting from platforms to AI chatbots.

“More and more people are going to chatbots to get information and these chatbots are probably not necessarily going to be citing news organizations,” said Sophia Crabbe-Field, lead author of the report compiled by the Centre for Media, Technology and Democracy at McGill University in Montreal.

“(AI firms) are becoming these huge, billion-dollar companies off of the information that comes from news.”

By interviewing media and tech experts and looking at other research and news reports, the McGill investigators concluded AI companies are scrambling for high-quality content to train their models, which has led them to news outlets and their content.

In most cases, news outlets never provided express permission for their output to be gobbled up by tech companies, pushing most publishers to want to want compensation for the work that helped train models.

Calculating appropriate compensation is tricky because publishers don’t always know how much of their content has been used by AI companies, nor how much those businesses are making off the models they trained on journalism.

“It’s hard to assign a clear dollar sign to news because it is a public good, so it can’t necessarily be something that we convert just into a dollar amount,” Crabbe-Field said.

Despite the difficulties, outlets are trying to assign value and extract compensation anyway.

News Corp., the publisher of the Wall Street Journal, Barron’s and the New York Post, signed a deal sharing news content with OpenAI last year. The deal allows the San Francisco-based creator of AI chatbot ChatGPT to train its models on News Corp. content. It is rumoured to be worth more than $250 million over five years.

The Associated Press, an international wire service, signed its own deal with OpenAI a few months later but hasn’t revealed the financial terms of the agreement.

Crabbe-Field’s report warns such deals undermine publishers in the long run. She suggests news outlets are better-off pursuing collective solutions rather than individual efforts.

“They should band together to potentially demand real concessions that help support the long-term viability of news, rather than making individual deals that provide immediate monetary value but maybe aren’t in the long term, in their best interest and that leave smaller or more diverse news organizations behind,” she said.

Asked about Crabbe-Field’s suggestion, industry group News Media appeared to agree that the deals being cut omit many of the publishers whose content is being used by tech firms.

“Most of the AI companies only seem interested in doing deals with one or two leading publishers in any market, yet we know they are omnivores who gorge on content from many sources, including smaller community outlets,” Paul Deegan, the group’s president, said in an email.

However, he saw the licensing agreements and recent lawsuits from “premium” publishers as establishing “guideposts” that give everyone a sense of what fair compensation looks like. He said they can eventually lead to a collective approach.

OpenAI was sued in November by a coalition of media companies claiming their copyright had been infringed.

Lawyers for the group, which includes The Canadian Press, Torstar, The Globe and Mail, Postmedia and CBC/Radio-Canada, said a possible compensation model could involve payments of $20,000 per work or an amount “the court considers just.”

It was followed in February by a lawsuit from the owner of the Toronto Star and U.S. publishers including Condé Nast, McClatchy, Forbes Media and Guardian News, which alleged Canadian AI company Cohere infringed on copyright by scraping copies of their articles from the internet without the publishers’ permission or compensation.

While Crabbe-Field’s report touched on such lawsuits, it was ultimately meant to examine the Online News Act. The federal legislation extracts compensation from some search engine and social media companies to help journalism businesses whose work has been repurposed by tech companies.

Google secured a five-year exemption from the Act by agreeing to pay $100 million per year to media organizations.

Meta, which owns Facebook and Instagram, is also subject to the law, but has avoided having to hand over money by blocking access to Canadian news on its platforms.

Crabbe-Field suggests future legislation could avoid giving Meta a way to skirt policies by containing “must carry provisions,” which would require platforms to continue hosting content from news publishers all covered by the law.

Crabbe-Field said such provisions are already imposed on cable and satellite providers who have to carry certain channels, like APTN, and if applied to social media platforms would have made Meta’s news ban more difficult to implement.

This report by The Canadian Press was first published April 13, 2025.

Tara Deschamps, The Canadian Press