by Artificial intelligence is rapidly changing how we browse the internet. Modern web browsers are no longer simple tools for opening web pages; they are evolving into AI-assisted platforms capable of summarizing content, predicting queries, and providing personalized recommendations in real time.
A prime example is Google’s integration of its Gemini AI into Chrome, the world’s most widely used web browser with a market share of roughly 70%. Gemini enhances Chrome with an AI assistant and an “AI mode” embedded directly into the address bar, allowing users to interact with content in more dynamic ways.
Beyond Google, nearly every major AI developer is seeking a presence in web search and browsing to maximize data collection. This trend has sparked a new wave of competition in the browser market, with companies like OpenAI and Perplexity reportedly eyeing acquisitions and partnerships to expand their AI capabilities.
The Data-Collection Power of AI Browsers
Unlike traditional browsers that mainly track cookies, search history, and cached data, AI-powered browsers gather much more granular information. These browsers monitor user behavior in real time, including how long someone spends on a page, which sections they highlight, what they copy and paste, and how they phrase prompts.
This data is invaluable for training large language models (LLMs). It allows AI developers to fine-tune search results, improve predictive algorithms, and deliver personalized advertisements. However, the depth of data collection raises serious privacy concerns. Information from emails, spreadsheets, financial records, and other personal content can potentially be exposed if not handled properly.
Karolis Kaciulis, a senior system engineer at Surfshark, highlights the stakes:
“My key question is, do companies see users’ privacy and security as their main focus area? Otherwise, what consequences are we, as users, going to face after one, five, or ten years of feeding agentic browsers with our most private and sensitive information?”
In essence, the more an AI browser understands a user, the more helpful it becomes—but the more sensitive data it collects.
How Much Data Do Popular AI Browsers Collect?
A closer look at some of the leading AI browsers shows significant differences in the type and amount of data collected:
Google Chrome (Gemini AI) collects 24 types of data, including names, location, device IDs, browsing history, search queries, product interactions, and purchase history.
Microsoft Edge (Copilot) gathers 12 data types, such as customer support info, browsing history, device ID, product interactions, diagnostic data, and multimedia files like photos or audio.
Perplexity (Comet), a browser with an integrated AI assistant, collects 10 types of user data, including location, device ID, purchase history, and diagnostic information.
Opera (Neon/Aria AI) is designed for privacy-conscious users. Its alpha version collects six anonymous data points, with no personally identifiable information, while the mobile version uses Aria AI for analytics in a similarly non-identifying manner.
Brave (Leo AI) takes privacy further, collecting only usage data for analytics and user ID for functionality. This makes Brave the most privacy-friendly AI browser among the ones analyzed.
SurfShark notes that even privacy-minded extensions like those for Firefox or standalone AI assistants may still transmit sensitive user data. Users need to understand these data collection practices to find a balance between functionality and privacy.
| AI Browser | Data Types Collected | Notes |
|---|---|---|
| Chrome (Gemini) | 24 | Includes behavioral and diagnostic data, purchase history |
| Edge (Copilot) | 12 | Includes multimedia, location, advertising info |
| Perplexity (Comet) | 10 | Includes device ID, purchase history, diagnostic data |
| Opera (Neon/Aria) | 6 | Anonymous data for analytics, no PII |
| Brave (Leo) | 2 | Usage data and user ID only, most privacy-focused |
AI Browser Extensions and Privacy Risks
A 2024 study by researchers from UCL, UC Davis, and Mediterranea University of Reggio Calabria analyzed ten popular AI browser extensions, including ChatGPT for Google, Merlin, and Microsoft Copilot. The study found that many extensions collect and share sensitive user data—sometimes including medical information, financial details, and educational records—without adequate safeguards.
Some AI assistants even transmitted full-page content from active tabs to their servers. Others, like Merlin, captured form inputs such as banking and health details. Extensions like Sider and TinaMind shared identifiable data such as IP addresses with analytics platforms, enabling cross-site tracking and targeted advertising.
These AI tools can also infer personal attributes such as age, gender, income, and interests, using this information to tailor responses across multiple sessions. Such practices violate privacy laws in the U.S., including HIPAA and FERPA, and may conflict with GDPR regulations in the EU.
Dr. Anna Maria Mandalari, lead author of the study, emphasized:
“While these AI assistants offer convenience, they often do so at the expense of user privacy. Many operate without transparency or proper consent, sometimes violating privacy legislation and terms of service.”
Who Owns Your Data?
At the heart of the privacy debate is ownership and consent. Users often unknowingly agree to share personal data to “improve AI accuracy” or “provide anonymous usage data.” In reality, anonymization can be weak, and behavioral patterns are often unique enough to identify individuals.
Tech companies benefit immensely from collecting detailed datasets, which can be monetized for advertising or used to advance proprietary AI models. Users, meanwhile, trade convenience for potential exposure of sensitive personal information.
Legal protections like GDPR and California’s CCPA give users some rights over their data, but these frameworks were established before AI-powered browsers became mainstream. Studies already indicate that many AI assistants operate in violation of these regulations.
The Takeaway
AI browsers are reshaping online experiences, providing smarter search, automatic summarization, and personalized assistance. However, these conveniences come with significant privacy trade-offs.
Users must consider how much personal information they are willing to share in exchange for AI-powered efficiency. Regulators and developers alike face the challenge of balancing innovation with robust privacy protections. As AI browsers evolve, the question remains: can we enjoy the benefits of intelligent browsing without sacrificing control over our most sensitive data?
FAQs
What private data does the Gemini AI browser collect?
Gemini collects 24 data types, including name, location, device ID, browsing and search history, product interactions, and purchase history.
How do agentic AI browsers impact privacy and security?
These browsers actively engage with content and track detailed user behavior. Sensitive data may be exposed, shared, or used for AI training.
What are the main risks of AI-powered browsers?
The biggest risks include surveillance, unauthorized data sharing, potential violations of privacy laws, and loss of control over sensitive personal information.
