AI Licensing
Somebody asked me this week what I thought about AMA and NEJM licensing to Open Evidence and earlier, Wolters Kluwer and LexisNexis to Harvey. My flip response, “if you can’t beat them, join them.” I wasn’t being facetious.
Publishers and aggregators of all stripes are looking at their businesses and considering how delivery of immediate answers with large collections of licensed and scraped data underpin or threaten their business and where new revenue can be garnered vs. where there is threat. Quite literally you name the firm and its offering, and we’ve been asked about it this past month and week. Our inquiry hotline has quite literally lit up.
Since GenAI crashed the party, we’ve been advising about AI Licensing and tracking licensing deals. In this blog from last year, we shared analysis about FT’s deal. Publishers must have ducks in a row to license their content; aggregators have to have more ducks in a row because securing publisher rights is one thing; but those publishers have to have their rights in order down the value chain. And who has the platform of choice in any ecosystem (up the value chain) is also an important consideration because only one or two platforms are going to be the platforms of choice when it’s all said and done. Those platforms increasingly look like big-tech.
So, while we’re answering the existential questions, we also spent time this week hosting non-competing general counsel, licensing execs, and other top officers in data, information, and b2b vertical tech firms with content and workflow solutions to discuss licensing to big tech in an off-the-record conversation about best practices and problem-solving. This only gets done by being in a physical or virtual room together, off-the-record and peer-to-peer. Add in Outsell expertise and research — alchemy happens.
Some takeaways:
- The landscape of AI intellectual property licensing has shifted rapidly from unlicensed web scraping to structured agreements that create value for both technology providers and publishers. Licensing reduces litigation risk while ensuring that models are trained on authoritative content, and it opens new monetization opportunities for content owners.
- Outsell has tracked and analyzed over seventy licensing deals with news and academic publishers among the earliest adopters. Wiley has signed agreements worth around $21 million, while HarperCollins pioneered a 50/50 author revenue split. Most deals are multi-year, non-exclusive, and cover a mix of text, images, video, and data.
- Regulation is evolving unevenly. The EU AI Act mandates transparency in training sources, strengthening publishers’ rights, while the UK has delayed legislation and the US is likely to regulate state by state. Court rulings remain mixed, sometimes favoring publishers and at other times AI firms.
- Big tech has shifted from scraping indiscriminately to pursuing targeted, licensed datasets, with attribution now central to their strategies. This is increasingly driven by engineers who seek specialized content — legal, financial, or video — rather than vast generic datasets. Evergreen materials are particularly valued.
- Success depends on being “AI-license ready.” This means auditing rights, consulting authors, defining clear licensing strategies, and promoting available content packages.
- Understanding the value-chain in which you operate is critical. See us about mapping the competitive ecosystem in which you complete and having clarity about vulnerabilities.
- Strong contracts must avoid perpetual rights, require attribution, prevent verbatim reproduction, and ensure audit rights.
The takeaway: without an AI licensing strategy content owners are leaving money on the table. Those that are transparent, rights-cleared, and proactive in packaging their content, are best positioned to command premium pricing and maximize value in the emerging AI economy. Just watch your back and make sure you’re clear where the new aggregators are coming from and which incumbents are threatened. It’s a wild, wild world and it’s moving fast!