OpenAI offers publishers up to $5M a year for training data
OpenAI is offering news publishers between $1 million and $5 million a year to license copyrighted articles for training its models, according to The Information. That falls short of Apple's reported multiyear offers topping $50 million. With web scrapers getting blocked and copyright lawsuits piling up, paying publishers directly is fast becoming the industry's cost of doing business.
The days of AI labs scraping the entire open web without consequence are officially winding down.
OpenAI is offering news publishers between $1 million and $5 million per year to license their copyrighted articles for AI training, according to a report from The Information. The figure offers one of the first concrete looks at the going rate for premium training data.
Apple bids higher while Big Tech sets precedents
OpenAI's reported offers look modest when stacked against the competition.
Recent reports suggest Apple is also courting major media organizations for AI training partnerships, putting at least $50 million on the table for multiyear deals. That dwarfs OpenAI's yearly ceiling for individual publishers.
OpenAI's numbers do line up with older media deals from social platforms. When Meta launched its now-discontinued Facebook News tab in Europe, it allegedly offered publishers up to $3 million annually for headlines and previews.
Yet tech platforms have paid far more when forced. Google committed $1 billion to news partnerships back in 2020. More recently, under pressure from local regulations, Google agreed to pay Canadian publishers around $100 million annually simply for linking to news stories.
Paywalls, blocked bots, and courtroom showdowns
OpenAI is coming to the bargaining table because scraping is no longer frictionless.
Major publishers including The New York Times and The Verge's parent company, Vox Media, have blocked OpenAI's web crawlers from touching their sites. Lawsuits have followed, with The New York Times suing OpenAI and Microsoft over claims that ChatGPT spits out its reporting nearly word for word.
Licensing deals are OpenAI's firewall against ongoing legal exposure. The company has already locked down agreements with The Associated Press and Axel Springer, the German media giant behind Politico and Business Insider, covering model training and newsroom tool development.
Google is also pitching newsrooms directly. The search giant reportedly demoed an AI tool dubbed Genesis to executives at The New York Times and The Wall Street Journal, aiming to ingest raw facts and write complete stories.
Why it matters
We think OpenAI's $1 million to $5 million pricing range is a clever attempt to anchor expectations low, but it feels fragile.
Offering single-digit millions helps OpenAI convert potential legal adversaries into commercial partners without blowing out its training budget. However, for storied newsrooms watching generative AI rephrase their hard-won scoops, that payout is pocket change compared to the existential threat AI poses to web traffic.
For publishers, the dilemma is stark. Taking the deal provides immediate non-advertising revenue, but signing off on training your replacement for a few million dollars a year is a hard pill to swallow. Holding out means funding grinding copyright litigation with no guaranteed victory.
Across the broader AI landscape, the era of zero-marginal-cost training data is dead. If deep-pocketed giants like Apple are willing to throw $50 million at content libraries, high-quality human data will quickly turn into an expensive bidding war that only the wealthiest tech firms can survive.
Sources (1)
- theverge.com(opens in a new tab)Originaltheverge.com