The most recent data-rights conflict between content owners and the AI industry is a lawsuit filed by social media giant Reddit against artificial intelligence company Perplexity, claiming it scraped and illegally used user posts to train its AI model.
The case submitted to the New York federal court on Wednesday also targeted three defendants, which Reddit says assisted Perplexity to gather its data: Lithuanian data scraper Oxylabs, “former Russian botnet” AWMProxy, and Texas start-up SerpApi.
Reddit claimed that the three smaller parties could steal its copyrighted material by “by masking their identities, hiding their locations and disguising their web scrapers as regular people.”
The allegations were denied by Perplexity, an AI-based search engine, which accused Reddit of “extortion” and fighting against an open internet, and the developer of the search engine, Serpapi, informed CNBC that it “strongly disagrees.” Therefore, Reddit claims and defends itself in court.
The lawsuit is among numerous similar lawsuits brought by content owners against AI companies that allegedly used copyrighted content without securing a license to train their large language models.
Reddit specifically has been at the forefront of that fight, having filed a similar ongoing suit against AI startup Anthropic in June.
In a statement, Chief Legal Officer of Reddit, Benjamin Lee, informed CNBC that AI companies are” locked in an arms race for quality human content” and that pressure has fueled an “industrial-scale ‘data laundering’ economy.”
Scrapers get around technology protection in order to steal data, which they can sell to their clients who want training content. Reddit is an ideal target as it is one of the biggest and most vibrant hubs of human communication ever developed.
In its lawsuit, Reddit, which is home to more than 100,000 communities in interest-based “subreddit” communities, stated that its user posts had become the most referenced source of AI-generated responses on Perplexity.
It stated it had served Perplexity with a cease-and-desist letter, and then it “increased the volume of citations to Reddit forty-fold.”
Previously, AI researchers have mentioned that the sheer number of moderated discussions on Reddit can be used to train AI chatbots to generate more natural-sounding responses.
Reddit, in the era of artificial intelligence, has endeavored to capitalize on its huge data pool, only giving it access to it based on licenses involving artificial intelligence agreements.
Such agreements have been signed by OpenAI and Google of Alphabet.
Perplexity, in a Reddit post in a reply to the lawsuit, claimed that it does not train AI models on content but just summarizes and quotes open Reddit discussions. Consequently, it claimed that it is “impossible” to sign a license agreement.
The statement indicated about the lawsuit that “A year ago, after explaining this, Reddit insisted we pay anyway, despite lawfully accessing Reddit data. Bowing to strong-arm tactics just isn’t how we do business. Show of force in Reddit’s training data negotiations with Google and OpenAI.”
Reddit stated that data licensing has been a growing source of revenue for the company. Reddit further added, “Perplexity believes this is a sad example of what happens when public data becomes a big part of a public company’s business model.”
The COO of Reddit, Jen Wong, informed the trade publication Adweek that AI licensing agreements with Google and OpenAI constituted about 10 percent of Reddit’s revenue in February.
