The NYT's three tests for an AI licensing deal

The New York Times' VP of strategic partnerships Adam Greenberg, a 14-year Google veteran, laid out the publisher's three-part test for any AI licensing deal at the Digiday Publishing Summit in Miami. The deal needs the right value exchange with express permission, sustainability beyond a one-time lump sum, and control and guardrails over where and how content is used. He called the emerging AI marketplaces "underdeveloped" and warned that a real-time marketplace valuing content for training or inference is years away. "It's very hard to get these things going. It takes years, and so a marketplace that is valuing content in real time for use in AI training or inference is just going to take a long time," Greenberg said. The Times' partnerships team has doubled to 10 people since he joined last summer, and its Amazon deal is held up as the one that clears all three bars. The uncomfortable reality he named is that few deals have been signed in the last few years because the conditions are hard to meet: "the devil is really in the details."

The crux

AI licensing is shifting from a one-off lump-sum windfall to an ongoing negotiation over value, attribution and control, and the publishers who treat it as a partnership rather than a payout are the ones getting leverage over linking and referral back to their sites.

Sell-side read

Copy the three-part test for your own deals, and refuse the lump-sum-only structure. The leverage in a licensing deal is not the cheque, it is the improved linking, attribution and referral mechanism that pulls users back to your own property.

Watch

Google's pay-per-value AI licensing program, which Digiday reports is rolling out to publishers, and whether it survives the same scrutiny Greenberg applies to Amazon.