The New York Times bans AI from scraping its stories

For the past year, tech giants have treated the open web like an all-you-can-eat buffet to feed their hungry AI models. Now, one of the biggest names in journalism is putting a lock on the pantry.

On August 3rd, The New York Times quietly overhauled its Terms of Service, explicitly banning the use of its content for training any machine learning or AI system without written permission, as reported by Adweek on August 10th.

Everything is off-limits

The Times did not leave any loopholes.

The updated terms cover text, photos, illustrations, video and audio clips, metadata, and even the publication’s signature "look and feel." The policy explicitly bars web crawlers from scraping this material to build "any software program, including, but not limited to, training a machine learning or artificial intelligence (AI) system."

Violators face unspecified fines or legal penalties. Curiously, as The Verge pointed out, the Times didn't immediately update its robots.txt file—the technical standard used to block automated crawlers. The legal fence is up, even if the digital gate was momentarily left unlatched.

The timing is hardly a coincidence. Google recently updated its privacy policy to state it could use public web data to train services like Bard, while OpenAI just gave publishers the option to block its GPTBot. Meanwhile, Microsoft updated its own rules to prevent others from using its outputs to train rival models.

Yet the Times isn't simply joining the crowd. While organizations like The Associated Press recently signed an open letter calling for collective industry protections, Semafor reported that the Times dropped out of a coalition attempting to negotiate jointly with tech platforms. That makes sense: back in February, the Times signed a $100 million, three-year partnership with Google covering distribution, marketing, and shared experimentation.

Why it matters

This isn't an outright refusal to work with AI. It’s an aggressive opening bid in a high-stakes business negotiation.

For publishers, the era of turning a blind eye to scrapers is officially dead. By updating its terms to include strict penalties and broad intellectual property definitions, the Times is laying the legal groundwork to demand licensing checks from anyone wanting access to its premium reporting.

For AI developers, the walls around high-quality training data just got significantly taller. If premier outlets refuse to be crawled for free and opt out of collective bargaining, only the best-funded labs will be able to afford legal, clean data agreements on a case-by-case basis.

We think the move is calculated and cutthroat. The Times knows its archival depth gives it unmatched leverage, and leaving the collective media alliance suggests it believes it can cut a far better deal on its own than standing shoulder-to-shoulder with smaller publishers.