Meta enters AI licensing fray, striking deals with People Inc., USA Today Co. and more

In the coming days, new and archival content from across People Inc.’s household brand titles, including People, Better Homes and Gardens, Allrecipes, Food and Wine, will be integrated into Meta’s LLM along with new and archival content from USA Today Co. and its network of 200 local publications and sports wires.

All companies are being tight-lipped on the terms of the deals, which remain undisclosed. It’s not clear whether they are lump-sum payments à la OpenAI’s preference or the pay-per-use style adopted by the likes of Microsoft.

Meta has trailed other AI companies’ push into licensing deals, with OpenAI, Amazon, and Microsoft having all gone all in. In contrast, Meta has the fewest, having partnered with Reuters last year to use its content to answer user questions in real time about news and current events. 

Now, that’s all changed, and it is a testament to just how competitive the LLM environment has become as AI companies race to own the most popular consumer LLM.

“To those who think the door for training data deals is shut, think again. Plenty of deals still to be had,” said a head of business development at a publisher, who spoke with Digiday under the condition of anonymity. The exec wondered if publishers’ recent moves to block AI crawlers from scraping their sites for content without compensation was beginning to show “green shoots of large models compelled to compensate publishers.”

The move signals that Meta is formalizing how it accesses news content for AI training and product development, and is striking given how the company has spent the past two years distancing itself from news distribution.

This year, publishers have reported a slight return to form in terms of Facebook sending them more referral traffic this year. But having been burned before, they know better than to treat that traffic as anything other than expected volatility.

Meta has radically reorganized its AI operations under a new division called Meta Superintelligence Labs, consolidating all its AI teams — from foundational model development to product engineering — under one roof. Publishers have previously told Digiday they believe this unit has signaled to the market a change in the company’s stance on accessing information for AI purposes. 

Over the past year, publishers have been coordinating through industry groups, including the IAB Tech Lab’s CoMP framework aimed at standardizing AI payments to publishers when their content is used for retrieval augmented generation (RAG) to synthesize answers to users’ prompts.

But nothing is tastier than a direct deal for individual publishers that can negotiate good terms. Danielle Coffey, CEO of News/Media Alliance, said deals like Meta’s mark a win for all publishers in the long-running battle to get paid for the use of their content by LLMs.

“These deals reinforce the strength of the licensing market and also demonstrate that there is value in our content and that licensing is, in fact, possible,” she said in an email statement. “Certainty in the law will help ensure there’s a fair exchange and that all rights are enforced.”

Three media execs said Meta’s deals were a good sign for the AI content licensing market.

A head of partnerships and business development at a publisher said it was “great news to see this change,” with Meta now showing that they are willing to pay publishers for AI content licensing, and this meant there was another player in the mix of large tech companies willing to open their wallets.

“Meta has tended to pay for news when legally required or when chasing new growth opportunities. This instance is probably a bit of both, and we’re encouraged by signs that Meta may be approaching this more responsibly,” said Digital Content Next CEO Jason Kint. “We hope it signals a growing recognition of the fair value of high-quality, trusted media.”

Luke Stillman, managing director at Madison and Wall, said the move is primarily about risk management and is consistent with Meta’s broader defensive strategy in terms of investing in areas that might threaten their current dominance.

“These licensing agreements preempt potential copyright litigation and help Meta show goodwill ahead of future regulatory scrutiny,” said Stillman. “And in the context of Meta’s broader AI investment cycle, it simply doesn’t make sense to spend billions on infrastructure but refuse to spend relative pennies on licensed content that can de-risk and accelerate future product development.”

From the publishers’ perspective, these deals are rational, he noted. “AI intermediaries pose an existential risk to many news and lifestyle brands; early licensing arrangements are likely to be the most lucrative as models continue to accumulate foundational knowledge; and publishers are better off taking the money while they still can, from as many platforms as possible, for as long as possible. Ideally, those revenues should help fund diversification into subscriptions, memberships, and events,” he said.