Today’s split screen is getting almost comedic: publishers are dragging OpenAI into sanctions territory over deleted logs, while the venture world keeps writing nine-figure checks to the same scraping infrastructure that makes those lawsuits inevitable. Nobody involved in the money seems to be waiting for the courts to decide who’s right.
Start with the legal front, because it’s the closest thing to a crisis this industry has. A 16-publisher coalition led by the Times says OpenAI hid its ability to search training data and may have destroyed 20 million ChatGPT logs — now seeking sanctions after a deposition blew up two years of claims that the logs were unsearchable, a story laid out in granular detail here and pressed again by the Daily News and others. If even a fraction of that holds up, it’s not a licensing dispute anymore, it’s a discovery-misconduct problem, and it should make every AI lab currently “negotiating” content deals nervous about what their own logs say.
That nervousness is spreading past the big mastheads. A Minnesota paper is publicly cheering the litigation on, and the owner of the Greenfield Recorder has filed its own suit. This is no longer a coastal-newsroom grievance — it’s a rural-paper grievance too, which is the kind of broad-based plaintiff class that tends to actually move legislation, not just headlines.
And yet: capital doesn’t care. Warburg Pincus just handed Oxylabs $130 million, minting a $3.6 billion scraping-infrastructure unicorn that Reddit has literally called a data “bank robber.” Bright Data, cut from the same web-scraping cloth, is now a strategic investor in a fincrime compliance startup’s $20 million seed round. Read those two together and the message is blunt: the market has priced in the litigation risk and decided it’s a rounding error next to the value of the pipes.
Mercor’s week tells the same story from the labeling side. Its CEO quietly funded, then bought a reinforcement-learning startup — a conflict of interest waved away as “the plan all along” — right as the company is reportedly doubling its valuation to $20 billion in a matter of months. Consolidating the training-data pipeline while self-dealing your way into it isn’t a great look, but investors clearly don’t need a great look, just a growth curve.
Elsewhere, the picture is less fraught but no less telling. Databento’s $97 million Series B is a bet that Wall Street’s market-data incumbents are expensive by design, not by necessity — a thesis I happen to agree with. DigitalOcean’s nine-figure AI contracts and 10x RPO guidance show compute demand is still climbing regardless of how the data underneath it was sourced. And in Brussels, regulators punted on social interoperability, proving once again that policy moves slower than either the lawsuits or the checkbooks.
Put it all together and the alt-data and training-data economy looks less like a market awaiting a legal verdict and more like one that’s already decided the verdict won’t matter much. Maybe it won’t. But someone’s going to be very surprised when a sanctions order lands on a company whose valuation assumed it never would.
Watch tomorrow for whether the SDNY judge actually grants sanctions against OpenAI — that ruling could reset the price everyone’s paying for scraped content, unicorns included.
Stories covered
- Small-Town Paper Cheers On AI Scraping Suits (Licensing & Legal)
- Fincrime Startup Tangos Banks $20M Seed, With Bright Data In Tow (Deals & Funding)
- Mercor’s CEO Invested in Deeptune, Then Bought It. He Says That Was Always the Plan. (Deals & Funding)
- DigitalOcean Lands Nine-Figure AI Deals, Talks Up 10x RPO Surge (Deals & Funding)
- NYT-Led Publishers Accuse OpenAI of Hiding Evidence, Seek Sanctions (Licensing & Legal)
- Another Local Paper’s Owner Sues Over AI Training Data (Licensing & Legal)
- Oxylabs Becomes a $3.6B Unicorn, Lawsuits and All (Deals & Funding)
- Mercor Reportedly Doubling Its Valuation to $20B in Months (Deals & Funding)
- NYT Accuses OpenAI of Faking Inability to Search Its Own Training Data (Licensing & Legal)
- EU Punts on Social Media Interoperability, Big Tech Wins Again (Licensing & Legal)
- Databento Raises $97 Million to Rewire How Wall Street Buys Market Data (Deals & Funding)
- Publishers Ask Judge to Sanction OpenAI Over Deleted ChatGPT Logs (Licensing & Legal)
