Perplexity's "Incognito Mode" is a "sham," lawsuit says

Lawsuit Storm: Perplexity's "Incognito Mode" Under Fire

Amid the rapid development of AI search and chat tools, a major lawsuit has struck the industry like a thunderbolt. As reported by Ars Technica on April 3, 2026, an anonymous plaintiff alleges that Perplexity AI's "Incognito Mode" is a pure "sham." The feature, intended to protect user privacy, is instead being used to cover up the illegal sharing of millions of chat records, funneling data to giants like Google and Meta to boost advertising revenue. The lawsuit has quickly sparked heated debate, exposing deep-seated privacy risks in the AI industry.

"Perplexity's Incognito Mode is not truly anonymous, but a guise for data capture. The company collaborates with Google and Meta to turn user queries into an advertising goldmine." — Excerpt from the lawsuit

Event Timeline: From Privacy Promise to Data Leak

Perplexity AI, as an emerging AI search engine known for real-time answers to user questions, has seen a surge in users since its launch in 2022. Its "Incognito Mode" was marketed as a privacy protection mode with "no tracking, no storage." When activated, chat records were supposedly not used for model training or ad tracking. However, through technical analysis and internal documents, the plaintiff claims that under this mode, user queries are still transmitted in real-time to third-party servers, including Google's ad platform and Meta's tracking system. The plaintiff asserts that in just the past year, over 5 million "incognito" chats have been shared, generating hundreds of millions of dollars in ad revenue sharing.

The lawsuit also names Google and Meta, alleging that they receive data via API interfaces for targeted advertising. For example, when a user searches for "best running shoes" on Perplexity, even with Incognito Mode enabled, their interest preferences are fed into Google's ad ecosystem to serve relevant product ads. This is not an isolated incident but a systemic design aimed at maximizing data monetization.

Industry Background: The Tip of the Iceberg in AI Privacy Crisis

Looking back at AI's development history, privacy issues have always been a shadow. OpenAI's ChatGPT faced criticism for storing user conversations by default, and its later "temporary chat" mode was still questioned. Google's Bard (now Gemini) and Meta's Llama series have faced similar allegations. Perplexity's uniqueness lies in its hybrid "search + generation" model, which relies on real-time web crawling but amplifies privacy risks. Under the EU's GDPR and U.S. CCPA, hiding data sharing is illegal, and this case could set a precedent.

Supplemental background: By 2025, the global AI data market has surpassed $100 billion, with advertising revenue accounting for over 40%. Although Perplexity claims to be "ad-free," it indirectly profits through data partnerships, making alliances with ad giants an industry norm. Similar incidents are frequent, such as in 2024 when Anthropic was exposed for sharing conversation data with Amazon to optimize Alexa.

Responses and Potential Impact

Perplexity quickly responded, stating, "Incognito Mode strictly adheres to privacy commitments. All allegations are misunderstandings, and we will fully defend ourselves." Google and Meta responded low-key, saying they "have not received formal notification and are investigating." Legal experts analyze that the plaintiff has a high probability of winning, which could result in fines of hundreds of millions of dollars for Perplexity and force the disclosure of data flows.

For the industry, this is a wake-up call. User privacy awareness is rising, and browser extensions like uBlock Origin have integrated AI tracking blocking. On the regulatory front, the U.S. FTC and the European Commission are drafting new AI laws requiring "privacy by default" rather than "opt-in."

Editor's Note: Privacy vs. Business—Who Prevails?

As an AI tech news editor, I believe the Perplexity incident reflects the double-edged sword of AI: behind innovation and convenience lies a data-hungry business logic. If the "Incognito Mode" is indeed a sham, it not only damages user trust but also shakes the foundation of the AI ecosystem. I recommend users enable VPNs, switch between multiple models, and consider open-source AI like Grok to diversify risks. In the long run, blockchain privacy technology may offer a solution, but legislative follow-up is necessary. AI prosperity should not come at the cost of privacy.

(Approx. 1050 words)

This article is compiled from Ars Technica, author: Ashley Belanger, date: 2026-04-03.