Social media platform Reddit sued the artificial intelligence firm Perplexity AI and three different entities on Wednesday, alleging their involvement in an “industrial-scale, illegal” financial system to “scrape” the feedback of thousands and thousands of Reddit customers for industrial acquire.
Reddit’s lawsuit in a New York federal courtroom takes purpose at San Francisco-based Perplexity, maker of an AI chatbot and “reply engine” that competes with Google, ChatGPT, and others in on-line search.
Additionally named within the lawsuit are Lithuanian data-scraping firm Oxylabs UAB, an internet area referred to as AWMProxy that Reddit describes as a “former Russian botnet,” and Texas-based startup SerpApi, which lists Perplexity as a buyer on its web site.
It’s the second such lawsuit from Reddit because it sued one other main AI firm, Anthropic, in June.
However the lawsuit filed Wednesday is completely different in the best way that it confronts not simply an AI firm however the lesser-known providers the AI business depends on to accumulate on-line writings wanted to coach AI chatbots.
“Scrapers bypass technological protections to steal information, then promote it to purchasers hungry for coaching materials. Reddit is a major goal as a result of it’s one of many largest and most dynamic collections of human dialog ever created,” stated Ben Lee, Reddit’s chief authorized officer, in an announcement Wednesday.
Perplexity stated it has not but acquired the lawsuit however “will all the time struggle vigorously for customers’ rights to freely and pretty entry public information. Our method stays principled and accountable as we offer factual solutions with correct AI, and we won’t tolerate threats in opposition to openness and the general public curiosity.”
SerpApi’s buyer success director, Ryan Schafer, stated in an e-mail: “We strongly disagree with Reddit’s allegations and intend to vigorously defend ourselves in courtroom.”
Oxylabs didn’t instantly reply to a request for remark Wednesday. AWMProxy couldn’t instantly be reached for remark.
Reddit compares the businesses it’s suing to “would-be financial institution robbers” who can’t get into the financial institution vault, so that they break into the armored truck as a substitute. The lawsuit alleges they’re evading Reddit’s personal anti-scraping measures whereas additionally ”circumventing Google’s controls and scraping Reddit content material straight from Google’s search engine outcomes.”
Lee stated that as a result of they’re unable to scrape Reddit straight, “they masks their identities, conceal their areas, and disguise their net scrapers to steal Reddit content material from Google Search. Perplexity is a keen buyer of a minimum of certainly one of these scrapers, selecting to purchase stolen information somewhat than enter right into a lawful settlement with Reddit itself.”
Reddit made the same argument in its lawsuit in opposition to Anthropic, alleging that the corporate ignored Reddit’s appeals to stop utilizing its content material. That case was initially filed in California Superior Court docket however was later moved to federal courtroom and has a listening to scheduled for January.
Together with digitized books and information articles, web sites corresponding to Wikipedia and Reddit are deep troves of written supplies that may assist train an AI assistant the patterns of human language.
Reddit has previously entered licensing agreements with Google, OpenAI, and different corporations which might be paying to have the ability to prepare their AI programs on the general public commentary of Reddit’s greater than 100 million day by day customers.
The licensing offers helped the 20-year-old on-line platform elevate cash forward of its Wall Road debut as a publicly traded firm final 12 months.
—Matt O’Brien, AP know-how author
