Perplexity AI Exposed: Your Private Searches Leaked to Meta

By the GPTAnon Editorial Team | April 2, 2026

---

You thought you were having a private conversation. You weren't.

On April 1, 2026 — and no, the irony of the date is not lost on us — a Utah man filed a class-action lawsuit in San Francisco federal court that should make every AI search user sit up and pay attention. The target: Perplexity AI, the darling of the "smarter search" movement. The allegation: that Perplexity has been secretly funneling your entire conversation history — every query, every follow-up, every deeply personal question — straight to Meta and Google through hidden tracking scripts.

Even when you thought you were browsing in "Incognito" mode. (We first covered Perplexity's data practices in our earlier investigation: Perplexity AI's Hidden Data Pipeline to Meta and Google.)

Let that sink in.

---

What the Perplexity Lawsuit Reveals About Your Data

The case — Doe v. Perplexity AI Inc., 3:26-cv-02803, filed in the U.S. District Court for the Northern District of California — paints a damning picture. According to the complaint, the moment you log into Perplexity's home page, tracking scripts are silently downloaded onto your device. These aren't garden-variety analytics cookies. The lawsuit alleges they give Meta and Google full access to everything you type into Perplexity's AI search — your queries, your follow-up questions, your entire conversational thread.

The lead plaintiff, identified only as "John Doe," described how he used Perplexity to discuss family financial information, tax obligations, investment portfolios, and financial strategies with the chatbot. Information he assumed was between him and the AI. Instead, according to the suit, that data was flowing directly to two of the largest advertising surveillance networks on the planet.

The lawsuit invokes violations of both the California Consumer Privacy Act (CCPA) and California's Electronic Communications Privacy Act (CalECPA), and seeks class-action status on behalf of affected California users.

> The core claim: Perplexity told users their conversations were private. They weren't. And the company's own privacy policy allegedly says nothing about these tracker deployments.

---

How Perplexity's Bot Disguises Itself as Chrome

Here's where it gets technically interesting — and more disturbing.

The complaint alleges that Perplexity's AI agent, called Comet, doesn't identify itself honestly when it crawls the web on your behalf. Instead, it reportedly transmits the exact same user-agent string used by Google Chrome. In plain English: when Perplexity's bot visits a website to answer your question, it pretends to be you, sitting in a normal Chrome browser.

Why does this matter? Because websites use user-agent strings to decide how to respond to visitors. A bot that announces itself as a bot can be blocked, rate-limited, or served different content. A bot that pretends to be Chrome? It slips right past those defenses. It's the digital equivalent of a process server wearing a pizza delivery uniform.

This isn't just a privacy issue — it's a deception issue. If Perplexity's agent is masquerading as a standard browser, then the data it collects and transmits exists in a gray zone where neither the user nor the websites being scraped understand what's actually happening.

---

Why "Incognito Mode" Won't Protect You From AI Tracking

Let's be brutally honest about something the industry doesn't want to say out loud: "Incognito mode" in an AI product means almost nothing.

When you open an incognito window in Chrome, all that does is prevent your browser from saving your local history and cookies after you close the tab. It does absolutely nothing to prevent the websites you visit — or the AI services you use — from tracking you on their end. Your ISP still sees your traffic. The service still logs your requests. And if that service has embedded third-party trackers, those trackers still fire.

A Stanford study found that all six major AI chatbots collect your data — and Perplexity is no exception. AI products have borrowed the language of browser privacy ("incognito," "private," "anonymous") without implementing the substance. When Perplexity offers an incognito-like mode, users reasonably assume their queries aren't being tracked or shared. The lawsuit alleges this assumption was not just wrong — it was deliberately cultivated.

This is a pattern we've seen before. Remember when "private browsing" first launched and everyone thought it made them invisible online? We're living through that exact same misunderstanding again, but this time the stakes are higher because the queries people send to AI are far more revealing than the URLs they visit. You don't Google your deepest anxieties. But you might ask an AI chatbot about them.

---

> "Private" AI search isn't private at all. GPTAnon never tracks your queries, never shares data with third parties, and never will →

5 Ways to Actually Protect Your AI Search Privacy

If you care about keeping your AI searches private, here's the uncomfortable truth: you need to assume that no commercial AI search product is truly private unless you can verify it yourself. Here's what that looks like in practice:

The stakes are higher than you think — researchers have shown that AI can now deanonymize you from your writing patterns alone. Here's how to fight back:

1. Use open-source, self-hosted AI models. Tools like Ollama let you run models like Llama, Mistral, or Phi locally on your own hardware. Your queries never leave your machine. This is the gold standard.

2. Route traffic through Tor or a trusted VPN. If you must use a cloud-based AI, at minimum obscure your IP. But understand this only hides who is asking — not what is being asked from the service itself.

3. Don't log in. Many AI products tie queries to accounts. If you must use a service like Perplexity, consider using it without creating an account.

4. Use browser-level tracker blocking. Extensions like uBlock Origin and browsers like Brave or Firefox with strict tracking protection can help catch third-party scripts before they fire — though this is an arms race.

5. Assume the worst. If a company's business model depends on advertising partnerships or data monetization, your "private" conversations are the product. Full stop. (This is exactly why GPTAnon exists — no ads, no tracking scripts, no third-party data sharing. Just AI that respects your privacy.)

---

Perplexity's Silence Speaks Volumes

When asked about the lawsuit, Perplexity spokesperson Jesse Dwyer offered a response that privacy advocates might charitably describe as "carefully worded": "We have not been served any lawsuit that matches this description so we are unable to verify its existence or claims."

Translation: we're not going to address the substance of these allegations until we absolutely have to.

This is boilerplate crisis communications, and it tells you nothing about whether the technical claims in the lawsuit are accurate. What it does tell you is that Perplexity isn't rushing to deny them.

---

The Bigger Privacy Crisis: Why This Matters for All AI Users

This lawsuit is a canary in the coal mine. The AI industry has spent the last three years telling users to "just trust us" with their most intimate questions — health concerns, financial worries, relationship problems, legal questions. And now we're learning that at least one major AI search company may have been treating those conversations as raw material for the advertising-industrial complex.

The privacy policies of most AI products are deliberately vague about third-party data sharing. The technical infrastructure of these products — built on web technologies, advertising SDKs, and analytics platforms — is inherently leaky. And the regulatory frameworks that are supposed to protect consumers are still playing catch-up with technology that moves at the speed of a venture capital funding round.

If the allegations in Doe v. Perplexity are proven true, this won't just be a story about one company's bad practices. It will be confirmation of what privacy advocates have been warning about since ChatGPT launched: the AI revolution was never going to respect your privacy (OpenAI's Atlas browser is another cautionary tale) unless it was forced to.

The question now is whether the courts — and users — will do the forcing.

---

---

Your AI conversations deserve real privacy — not privacy theater. GPTAnon connects you to 25+ AI models including GPT-5, Claude, Gemini, and DeepSeek. No tracking scripts. No data shared with Meta or Google. No account required. Try genuinely private AI →