Photo by Ray Harrington on Unsplash
Publishers are treating AI content deals as a profitable baseline, when they’re actually outliers according to Scott Purcell, co-founder of independent publisher, Man of Many.
In an interview with AdNews, Purcell outlined how AI content deals between publishers and large tech giants were a losing gamble for the industry.
“Anyone who's talking about AI content licensing to save publishers is absolutely kidding themselves.”
He pointed to News Corp's reported content licensing agreement with OpenAI as the exception rather than the rule.
“The deal itself is good business and credit to News Corp for getting it done,” said Purcell.
“The harm is in how the rest of the industry reads it. It's reportedly up to $250 million over five years, part cash and part OpenAI technology credits, for a portfolio that includes the Wall Street Journal, The Times and the Australian mastheads.
“Since 2023 the major labs have committed roughly $3 billion across 50-odd publisher deals by one count, and it's overwhelmingly concentrated in a few dozen global brands.
“There is no queue forming for everyone else. Scale the public benchmarks down to a mid-size independent publisher and you get low hundreds of thousands a year at best, and the AI labs aren't calling. “
“So yes, boards are treating an outlier as the market rate, and that false assurance is delaying the harder strategic conversation most publishers need to have.”
Purcell explained the economics become even starker when smaller publishers calculate what their archives are actually worth in the current licensing market.
“There's no clever modelling behind it, which is exactly the point. Bright Data lists scraped news article datasets on the Snowflake Marketplace starting at USD $250 per 100,000 records,” he said.
“That is a quarter of a cent per article.
“Man of Many has published more than 10,000 articles over 13 years, so at the listed market rate set by Bright Data our entire archive prices at proxy of about $25.
“Those articles cost us millions in salaries, photography, travel and product testing to produce.”
Seeing that number made the Man of Many co-founder reconsider how he viewed AI giants' valuation of news.
“Seeing it written down told me the licensing market doesn't price content on what it cost to make or what it's worth inside a model.
“ It prices it against the next-cheapest substitute, and the substitute is industrial scraping that US courts have found legal.
“Meta and X both sued Bright Data over scraping public data, and both lost in 2024. When the substitute costs a quarter of a cent, that's the floor every negotiation starts from.”
Looking to the future, Purcell believed what’s done is done for existing archives, but that publishers needed to do more to protect their output moving forward.
“You have to separate the archive from the publisher. The archive as a static commercial asset is done: it has already been scraped, and you can't un-scrape it,” he warned.
“Cloudflare's data now puts more than half of all web traffic as non-human, and 52 per cent of crawler requests are serving AI training, up from 22 per cent a year earlier.
“The library has been read. What still commands money is what a scraper can't already have: today's content, provenance, legal certainty, and work, data and research that exists nowhere else.
“So there is absolutely a future, but it belongs to publishers as ongoing producers, not as archive owners.
“We've structured our own licensing that way, recurring and usage-based rather than one-off archive buyouts, including a 50 per cent revenue-share arrangement with ProRata's Gist.ai and bot-level measurement through TollBit.”
As for who is best positioned for the AI era, Purcell said original journalism would determine the winners.
“I won't hand out report cards on my peers, but the criteria matter more than the names,” said Purcell.
“The right side of the line is genuine original reporting, first-party reviews and photography, real-time coverage, and niche or local journalism the wires don't touch.
“News Corp and Nine have both the leverage and genuine original output. The trade press, frankly including AdNews, is a good example of niche original reporting a scraper can't find twice.
“The wrong side is any publisher whose output is AI Slop content, a rewrite of the same press release, announcement or Reddit thread as everyone else.
“That content is already priced, and the price is a quarter of a cent.”
Given uniqueness is the premium market, Purcell identified willingness to change content output as the future for survival.
“Commodity back catalogues are priced, and the price is sitting on the Snowflake Marketplace for anyone to see.
“But the forward fight is open to anyone willing to change what they make.
“Uniqueness is not a moonshot. It's original photography, first-hand testing, named experts and reporting that doesn't exist anywhere else. That is just journalism, which is what the industry was meant to be selling all along.
“The scrap economy is killing the copy-paste publishing model, but zero-click search was already doing so.
“So I'd flip the framing: the majority of the industry hasn't lost the fight so much as lost a business model that was ending anyway.
“The publishers who move to work that a scraper can't find twice will be fine. The ones selling $25 archives and calling it an AI strategy won't."
Purcell also believes advertisers are currently subsidising AI development.
“Nearly everything in the training corpus was funded by advertising sold on the promise of human attention,” he said.
“Advertisers paid for the journalism, the journalism trained the models, and the models now answer the questions that used to deliver those advertisers an audience.
“With more than half of web traffic now non-human, a growing share of the pages advertisers underwrite are being read by machines that will never buy anything.
“Nobody signed up for that knowingly, but that is the value chain as it stands.”
Publishers will survive by owning audience and licensing structures, according to Purcell.
“Survival looks like owned audience through newsletters and direct channels, commerce and events revenue, video, podcast and audio formats that are harder to substitute, and licensing structured as recurring revenue share rather than one-time buyouts.
“Blocking alone is not a strategy either: one analysis last year found publishers who blanket-blocked AI crawlers saw around a 23 per cent decline in visits, while non-compliant scrapers ignored the block anyway.
“The play is selective. Block pure training scrapes, allow retrieval with attribution, and build the licensing leverage you can't create alone through collective arrangements.
“By accepting that the low quality, clickbait, social media pageviews, and traffic economy that built the digital publishing industry in the 2010s is not coming back,” he said.
“News Publisher clicks have reportedly fallen 50 per cent year on year, roughly 60 per cent of searches now end without a click, and Facebook referrals have halved.”
Have something to say on this? Share your views in the comments section below. Or if you have a news story or tip-off, drop us a line at adnews@yaffa.com.au
Sign up to the AdNews newsletter, like us on Facebook or follow us on Twitter for breaking stories and campaigns throughout the day.