In a dispute that underscores the growing friction between artificial intelligence companies and content platforms, Perplexity — a fast-rising AI “answer engine” founded by Aravind Srinivas — has rejected allegations that it unlawfully scraped Reddit data for profit.
The lawsuit, filed on October 22 in the U.S. District Court for the Southern District of New York, accuses Perplexity and three other entities — Oxylabs UAB, AWMProxy, and Texas-based SerpApi — of “industrial-scale, unlawful” scraping of Reddit user content. Reddit claims the companies extracted millions of comments without authorization, monetizing user-generated discussions to strengthen AI products and search tools.
Perplexity has dismissed the claims as “baseless,” framing the lawsuit as part of a broader struggle over who controls public internet data in the AI era.
Perplexity’s Defense: Public Data, Public Rights
In a sharply worded public post, Perplexity characterized Reddit’s lawsuit as a “sad example” of overreach, arguing that the platform was using litigation as leverage in its negotiations with larger AI firms such as Google and OpenAI.
FCRF Launches CCLP Program to Train India’s Next Generation of Cyber Law Practitioners
Perplexity emphasized that it does not “train foundation models,” describing itself instead as an “application-layer company” that merely aggregates and summarizes public discussions. “We lawfully access publicly available Reddit data,” the company wrote. “We will play fair — but we won’t cave.”
The firm also claimed that Reddit had “insisted on payment” despite Perplexity’s refusal to license content it says it doesn’t use for training. Calling Reddit’s demand a “strong-arm tactic,” the company alleged that the social media giant was “trying to extort smaller firms” while cutting exclusive data deals with its larger rivals.
The Broader Stakes: Licensing, Power, and Precedent
The lawsuit is not Reddit’s first. In June, it sued AI startup Anthropic on similar grounds, accusing it of using Reddit posts to train large language models without a license. But the latest case is more expansive — targeting not just a single AI company but also the data-sourcing intermediaries that fuel the industry’s training pipelines.
Reddit’s complaint marks a new front in the industry’s evolving legal and ethical battle over AI training data. The platform, which went public last year, has positioned content licensing as a key source of post-IPO revenue. Its deals with OpenAI and Google reportedly run into tens of millions of dollars, granting those firms access to Reddit’s vast corpus of user discussions for AI training.
As model makers pull back from such agreements amid rising costs and public backlash, Reddit’s lawsuit against Perplexity and its data suppliers could set an important precedent for how “public data” is defined and monetized in the age of generative AI.
AI Ethics and the Future of Public Data Access
At the heart of the dispute lies a philosophical and legal tension: where does the “public” nature of online data end and proprietary ownership begin?
Reddit argues that even publicly visible content remains under its copyright and user agreement protections, and cannot be used for commercial AI purposes without consent or payment. Perplexity and other AI companies counter that restricting access to publicly available information would stifle innovation and consolidate data power in the hands of tech giants able to afford expensive licensing deals.
For now, the case has triggered broader scrutiny of how AI startups build their systems — not only by governments and courts but by the very platforms on which much of the modern internet’s collective knowledge resides.
As both sides prepare for what could be a landmark battle, the Reddit–Perplexity standoff has become emblematic of a deeper question for the digital age: who truly owns the data that defines our shared online world?
