Perplexity

Srinivas put the platform at 780 million queries in May 2025, growing more than 20% month over month, against 3,000 queries on its first day in 2022^[2]. Valuation followed the same curve, from roughly 1 billion dollars in April 2024 to 14 billion in June 2025, 20 billion that September and about 21.21 billion in early 2026^[5].

A Tow Center audit published in Columbia Journalism Review in March 2025 ran 1,600 queries across eight AI search tools, each asking the tool to identify the source of a news excerpt. Perplexity answered 37% of them incorrectly, against more than 60% across the eight tools together^[6]. The inline citations are therefore a claim about provenance rather than a verified one. See citation accuracy.

In June 2024 Wired and the developer Robb Knight documented Perplexity fetching pages from undisclosed IP addresses behind spoofed user-agent strings, reaching content its declared crawler had been blocked from^[7]. On 4 August 2025 Cloudflare published a wider finding: stealth, undeclared crawlers across tens of thousands of domains and millions of requests a day, rotating IP addresses and network numbers and, once blocked, disguising the traffic as a Chrome browser on macOS^[3]. Cloudflare de-listed Perplexity as a verified bot and added blocking heuristics to its managed rules^[8]. Perplexity disputed the account, arguing that an agent answering one person's question is not a crawler that systematically visits millions of pages to build a database^[9]. See AI crawlers.

Dow Jones and the New York Post, both News Corp properties, sued Perplexity in the Southern District of New York for copyright infringement, false designation of origin and trademark dilution, arguing the product let readers skip the links to the publishers' own sites^[4]. Reddit filed its own scraping suit in October 2025^[10].