Perplexity built its reputation on a simple idea: AI-powered search that gives you direct answers without surveillance. No ad profile. No tracking cookies trading your clicks to a dozen marketing platforms. Just an answer to your question.
Then Perplexity launched Comet.
Comet is a browser — not just a search tool. And a browser that knows everything you do online is a fundamentally different product from a search tool that knows what you searched for. The privacy story changes considerably once you understand what Comet collects and why.
What Comet Actually Is
Comet launched in mid-2025 as Perplexity’s entry into the browser market. The pitch was an AI-native browser: instead of searching within a browser, the browser itself becomes an AI assistant that has full context about your online activity.
The value proposition is real. If your browser knows you’ve been researching mortgage rates, reading about specific neighborhoods, and comparing schools in different districts, it can answer “what’s the best neighborhood for us?” with genuine context. That requires the browser to read everything you do — not just your search queries.
This is the trade. The AI assistant becomes more useful exactly as it collects more data. The scope of collection is the product, not a side effect.
What Comet Collects
Perplexity’s privacy documentation for Comet describes collection of browsing history, the content of web pages you visit, and form data — which can include information you type into websites.
The “form data” category is the most sensitive. Forms on websites capture:
- Login credentials (username fields, email fields)
- Passwords (though browsers typically handle these separately)
- Payment information
- Personal details you enter during account creation or checkout
- Health information on medical or insurance sites
- Messages you type before sending
The browser processes this data to give the AI assistant context. Where that data is transmitted, stored, and retained is the question that determines the privacy risk.
Perplexity’s data handling sits under US jurisdiction, which means it’s subject to US discovery rules. Legal demands for your browsing history, submitted forms, and web activity are served on Perplexity — not on you. You may not be notified when they comply.
The Difference from Perplexity’s Search Product
Perplexity’s core search product has a relatively contained privacy footprint. You submit a query, the system retrieves and synthesizes information, and the interaction is primarily one-directional.
Perplexity search still collects query data and can be associated with an account if you’re logged in. But the scope is bounded by what you actually search for.
Comet is categorically different. A browser sees:
- Every website you visit, including ones you’d never search for in a search engine
- The content on those pages
- How long you spend there
- What you click
- What you type
This is why browsers are among the most sensitive applications on any device. The browser has more insight into your actual behavior than any search engine, because most of your online activity never starts with a search query.
When Perplexity extends its AI product into a browser, it’s acquiring access to that entire behavioral record — not just your Perplexity queries.
The Monetization Question
Perplexity is a venture-backed company with significant investment but, like most AI companies, a path to profitability that remains under construction.
The company has disclosed that it will introduce advertising in some form. Search advertising is a known quantity. Browser-level behavioral data — the full picture of online behavior — is more valuable to advertisers than search queries alone, because it captures intent and behavior across the entire web.
The advertising model for browsers has been established by Chrome. Google’s browser collects activity data that informs Google’s advertising business. The browser is an audience intelligence product as much as it is a utility.
Whether Perplexity’s eventual monetization model for Comet follows this path is not definitively established. But the infrastructure for that model — comprehensive browsing data collection — is built into the product.
How This Differs from Other AI Search Tools
It’s worth situating Comet within the broader landscape of AI-powered search and browsers:
ChatGPT Search. OpenAI’s search feature operates within the ChatGPT product. It collects query data and conversation history. It does not observe your entire browser session.
Google AI Overviews. Google’s AI search integration exists within Chrome and Google Search. Google already has comprehensive behavioral data from Chrome; AI Overviews adds synthesis capability on top of existing collection. If you’re in the Google ecosystem, the additional privacy cost of AI Overviews is relatively small because Google already knows what you do in Chrome.
Brave Leo. Brave’s AI assistant runs with strong default privacy protections and processes queries without sending them to remote servers by default. Brave Leo is designed specifically to avoid the data-collection model that powers most AI assistants.
Arc and other AI browsers. Newer browsers with AI integration have varied approaches to privacy. The common thread is that browsers with AI features need more data to make those features useful — which creates a fundamental tension with privacy-first design.
Comet sits at the high-collection end of this spectrum. Its AI assistant is designed to be more capable precisely because it has more context. More context means more collection.
What the Perplexity Track Record Suggests
Before Comet, Perplexity attracted criticism in 2024 for its web crawling practices — specifically, the Perplexity crawler was found to be impersonating the user agent strings of other crawlers to bypass websites that had blocked Perplexity. This was reported and discussed extensively on Hacker News and in tech media.
The details matter less than the signal: Perplexity has demonstrated a willingness to operate in ways that push against published norms when it benefits their product. A browser from a company with that track record, with access to your full browsing session, warrants careful evaluation.
This doesn’t mean Comet is malicious. It means the privacy representations in the terms of service deserve scrutiny, and “we value your privacy” from Perplexity should be evaluated against their operational history, not just their marketing copy.
Practical Guidance: How to Use Perplexity Privately
If Perplexity’s search product is useful to you, the privacy-conscious approach is to use it in a browser where it has limited context:
Use Perplexity in a browser that minimizes data collection. Firefox with uBlock Origin, Brave, or Safari with tracking prevention enabled limits Perplexity’s ability to observe behavior outside their own product.
Avoid logging into Perplexity across devices if you prefer query history not be associated with your identity. Logged-out queries are typically processed with less persistent association.
Don’t use Comet as your primary browser. If the feature that interests you is AI-powered search answers, you get that from Perplexity’s web product. The browser-level features require browser-level data collection; if you don’t need those features, you don’t need to pay that price.
Read Comet’s privacy policy specifically. Perplexity’s privacy policy for its search product may differ from Comet’s data collection terms. Before installing any browser, read what data it collects, where it sends that data, and how long it retains it.
The Broader Pattern in AI Products
Comet represents a general dynamic in the AI industry: the path to more capable AI products goes through more comprehensive data collection.
AI assistants that know your context are more useful than AI assistants that don’t. The more context they have, the more useful they become. The question — and it’s genuinely hard — is whether users get enough utility from that context to justify the surveillance it requires.
The answer is personal and depends on what you value, what you’re doing online, and how much you trust the company doing the collecting. But it’s worth making that tradeoff consciously, rather than downloading a browser because it has good AI features without thinking through what “good AI features” requires.
Why This Matters for Personal Files
If you use cloud storage, a browser with full context access can observe when you’re accessing your storage service, what files you open, and what you do with them online.
This is different from the storage service itself having access to your files. It’s the browser — sitting above every application you use — observing your interactions with everything.
Personal file storage that sits behind good encryption practices reduces the blast radius of browser-level data collection. The browser might observe that you accessed your storage service, but it shouldn’t be able to read the contents of encrypted files.
daftei stores files with AES-256 encryption at rest and TLS 1.3 in transit. It doesn’t expose file contents in ways that would be observable by a browser’s form-capture or page-reading functions.
That layer matters when the browser itself is a surveillance product. The storage layer should be the one part of the stack that isn’t.
The Core Question
Perplexity search built trust by being simpler and less intrusive than Google. Comet grows from a different seed: it’s more useful than a search tool because it knows more than a search tool knows. That’s the product. The data collection isn’t a cost of the product; it’s how the product works.
Whether that’s a trade worth making is a question only you can answer. But you should answer it explicitly — not discover the terms later.