Brave Search API lawsuit: what it means for AI data buyers

0:05Host:MBA Training. Ten minutes on Brave lawsuit puts your AI data supply chain in the dock. All but four words. The Times ran a test on Brave's API and got its own article back, minus four words.

0:19Expert:And it wasn't just any article. It was the robot theme park piece. PPC Land has the headline: "Mixed Human Feelings After a Day at a Park Full of Robots."

0:31Host:So a machine copied a story about how people feel about machines.

0:36Expert:The complaint writes itself. Well, the lawyers wrote it, but you get the idea.

0:41Host:Okay, set it up. Who's suing who?

0:44Expert:On November 2nd The New York Times sued Brave Software in Manhattan federal court. A few hours later the Daily News, the Chicago Tribune and the Orlando Sentinel filed their own suit. The Next Web covered both.

0:59Host:Brave. The privacy browser. The ad blocker people.

1:03Expert:Same company. It also runs its own search engine, and behind that sits the Brave Index. The complaints put it at more than 30 billion pages.

1:14Host:Wait, what? Brave has its own index? I assumed everyone just rents Google's.

1:19Expert:Most do, and that's Brave's pitch. It sells access to the index through a Search API, a paid pipe your software calls to get results back.

1:29Host:And who's on the other end of the pipe?

1:31Expert:The complaints name Anthropic, Perplexity, Mistral, xAI and Cohere as customers. None of them is a defendant.

1:40Host:Oh, come on. The buyers walk?

1:43Expert:For now. PPC Land also points out that the filings don't say which of those firms use the API today, or whether any of them actually got Times articles.

1:55Host:So the papers went after the...

1:57Expert:...the middleman. Yes. The Tribune papers' lawyer, Steven Lieberman, told ShiftDelete: "Buying stolen goods secondhand doesn't make them legitimate."

2:07Host:And the complaint calls Brave a burglar and a fence in the same breath.

2:12Expert:That's the three newspapers' complaint, quoted by The Next Web. Burglar for the scraping, fence for the reselling.

2:20Host:Here's my problem with all this. Every data team I know buys web data the way they buy printer toner. Compare the price, compare the speed, click subscribe.

2:29Expert:And the price was right there. PPC Land had the Search API at five dollars per thousand requests back in July.

2:37Host:Cheap. Provenance not included.

2:40Expert:That's the awkward part. Teams building RAG, retrieval-augmented generation, where the model pulls documents in at question time to ground its answer, treat the search feed as plumbing.

2:53Host:Plumbing nobody inspects until the basement floods.

2:56Expert:And the publishers say the pipe was full of their content from day one.

3:01Host:Let's talk about the crawler. The bot that actually reads the pages.

3:05Expert:This is the part I find hardest to defend. The publishers say Brave has never disclosed what its crawler is called. If a bot never announces itself, a site owner can't block it.

3:17Host:The crawler with no name. It's a spaghetti western. Clint Eastwood squinting at your robots.txt.

3:24Expert:robots.txt being the little file where a site tells bots what they can and can't read. Hard to obey a sign that's addressed to somebody else.

3:35Host:Did anyone just... ask them to stop?

3:38Expert:The Times says it did. Its spokesperson, Graham James, told Techaeris that Brave ignored repeated requests to stop.

3:46Host:Ghosting as a legal strategy. Bold.

3:49Expert:Bold until it shows up as an exhibit.

3:52Host:Okay, give me Brave's side, because I'm already half convinced they're toast.

3:57Expert:Brave hasn't said anything publicly. CyberInsider and Techaeris both noted it hadn't responded. But its position is on the record. In March 2025 it sued News Corp, asking a court to declare that web indexing is fair use, the American doctrine that allows some copying without a licence.

4:18Host:They sued first? Cheeky.

4:20Expert:Pre-emptive. News Corp countersued and accused Brave of theft. And the argument isn't crazy. Every search engine copies pages to build an index. That's how search has always worked.

4:33Host:Sure, but indexing so you can point me to the article is one thing. Handing a chatbot the article minus four words is something else.

4:42Expert:I'm with you on that, for what it's worth. My bet is the court looks at each stage separately. The index might survive. Full text piped into AI products looks much harder to defend.

4:54Host:Is that a fact or a vibe?

4:57Expert:A vibe. A well-read vibe. Nobody has ruled.

5:01Host:Then I'll be the unpopular one. I think the buyers belong in the dock. Anthropic, Perplexity, all of them. You don't get to say "we just bought the feed."

5:12Expert:I don't buy that. Not yet.

5:14Host:Why not?

5:15Expert:Because the evidence against them in these filings is thin. The Next Web says the complaints cite a Brave blog post saying Anthropic used the API to train Claude 2. That's Brave's own marketing, from 2023. It tells you nothing about what anyone pulled last month.

5:35Host:So the best exhibit against Anthropic is Brave bragging.

5:39Expert:More or less. And if the rule becomes "every buyer answers for every supplier's crawl," then every company with web search in a chatbot is a co-defendant in waiting.

5:51Host:Good. Maybe then they'd finally ask where their data comes from.

5:55Expert:Or they'd all run back to Google and Bing, the only other independent indexes by Brave's own description, and pay whatever those two charge. A market with two sellers doesn't do publishers any favours either.

6:09Host:I still think "we didn't know" is a weak defence for anyone with a data budget.

6:14Expert:And I think a lawsuit against a supplier is the wrong place to settle that.

6:18Host:Then we leave it there. Badly.

6:21Expert:Our natural state.

6:23Host:What about Perplexity? They're on the customer list, and the Times is already suing them.

6:28Expert:They didn't keep quiet. According to Techaeris, their spokesperson said: "The New York Times is overpaying its attorneys and underpaying its journalists."

6:38Host:Oof.

6:39Expert:ShiftDelete says they also called the filing "disinformation," without pointing to any specific errors.

6:46Host:That one gets to me a bit. Somebody spent a day in a robot park and wrote the piece, and now the argument is about legal fees. The reporter is the one person here who isn't getting paid twice.

6:58Expert:Fair. And that's the publishers' money argument. The Next Web says the newspapers cite a study finding that AI crawlers send far less referral traffic than Google search does. Fewer clicks means fewer subscriptions.

7:13Host:What else is in there? The paywall stuff?

7:16Expert:Yes, the suits go beyond the API. They also target a browser setting that gets around paywalls, and Brave blocking publishers' ads while selling its own. That's where the DMCA claims come in. The DMCA is the US copyright law that bans getting around access controls like paywalls.

7:36Host:An ad blocker accused of blocking ads. Stop the presses.

7:40Expert:Their presses, specifically.

7:43Host:How big is the exposure?

7:45Expert:The Tribune-owned papers say more than two million registered articles are at stake in their case, according to ShiftDelete.

7:52Host:And how big is Brave?

7:54Expert:The newspapers' complaint puts its revenue at over a hundred million dollars for 2025, per The Next Web. This is a real company, not some scraper in a garage.

8:05Host:So what changes for a data team calling a search API inside an agent? By agent I mean a model that goes off and fetches things and acts on its own.

8:15Expert:Legally, today, nothing. But the procurement conversation shifts. It used to be latency, coverage and unit price. Now it's where the crawl came from, and whether the vendor will stand behind it in court.

8:30Host:Indemnity. The clause where the vendor promises to cover you if their data gets you sued, and suddenly nobody remembers how to spell it.

8:39Expert:Nobody enjoys that conversation. But a feed that could get enjoined, meaning a court orders it stopped, is a dependency risk, same as a cloud region going down.

8:51Host:Except a cloud region doesn't get sued by the Chicago Tribune.

8:55Expert:Not yet.

8:56Host:So what are you watching?

8:57Expert:Brave's formal answer first. It will show whether they defend the whole stack or try to wall the index off from the AI products. Then the News Corp case, which could get to a fair use ruling sooner.

9:10Host:And whether the five buyers stay bystanders.

9:13Expert:That's the big one. If a judge decides the API was the delivery van for stolen goods, the people who signed for the parcels start sweating.

9:23Host:Meanwhile the crawler with no name is still riding through town.

9:27Expert:Reading everyone's signs and obeying none of them.

9:31Host:What I can't shake is this. If the index survives and the answers don't, who gets to decide where search stops and an AI answer starts: the judge, the vendor, or the people buying the feed? Sources: The Next Web, BetaNews and PPC Land. Every link is in the show notes. We'll leave it there. The CDO self-assessment at mba-training.com will tell you where you really stand.