📰 AI News: Featured Chrome VPN Caught Intercepting Millions Of AI Chats

📝 TL;DR

A massively popular Chrome VPN extension with a “Featured” badge was quietly logging people’s AI chats and sending them to its own servers. If you are using a browser VPN or “AI protection” extension, this is your reminder to check what you have installed.

🧠 Overview

A Chrome extension with around six million users and an official “Featured” badge was found intercepting every prompt and response from major AI chatbots, including ChatGPT, Claude, Copilot, Gemini, DeepSeek, Grok, Meta AI, and Perplexity.

The data was captured in the browser and forwarded to servers controlled by the extension’s publisher and an associated data company. This is less about a single bad extension and more about how easily trust in the browser extension ecosystem can be abused.

📜 The Announcement

On December 15, 2025, security researchers revealed that a popular VPN browser extension turned into an AI data vacuum after a July 9 update.

The extension, promoted as a “secure free VPN” and carrying a “Featured” badge in the Chrome Web Store, added code to silently monitor AI chat pages and harvest conversations by default.

The same AI chat harvesting behavior was also spotted in three related extensions across Chrome and Edge, pushing the total affected install base to over eight million users.

⚙️ How It Works

• Targeted AI sites - The extension ships special scripts for each major AI chatbot, for example for ChatGPT, Claude, Gemini and others, which activate whenever you visit those sites.

• Browser API hijack - The script overrides key browser network functions like fetch and XMLHttpRequest so every AI request goes through the extension first.

• Full conversation capture - It collects your prompts, the AI responses, conversation IDs, timestamps, session metadata, and which AI platform or model you are using.

• Silent data exfiltration - That data is then sent to remote analytics servers controlled by the extension operator and shared with an affiliated ad intelligence and brand monitoring company.

• Privacy policy loophole - The updated privacy policy says AI prompts and outputs are collected for “safety” and “analytics” and will be anonymized, while also admitting that sensitive personal data may still slip through.

• Featured badge effect - Multiple extensions from the same publisher, including VPNs, “browser guards,” and ad blockers, all carried “Featured” badges in Chrome or Edge, which made them look more trustworthy than they actually were.

💡 Why This Matters

• Featured is not the same as safe - A platform “Featured” badge feels like a trust stamp, but this case shows it is not a guarantee that an extension respects your privacy.

• AI chats are not harmless small talk - People share health worries, financial details, product plans, and personal stories with AI tools, so leaking those chats is closer to leaking private journal entries than generic browsing data.

• Auto updates cut both ways - Automatic extension updates are convenient, but they also mean an extension can change behavior overnight without you noticing until your data has already been collected.

• Data brokers love this kind of feed - Captured AI conversations can be turned into highly detailed behavioral and intent profiles that are extremely valuable for advertisers, and very uncomfortable for users.

• Trust in “AI helpers” is fragile - As more AI themed extensions pop up, this kind of incident can make people hesitant to experiment, which slows down adoption and makes AI feel risky instead of empowering.

🏢 What This Means for Businesses

• Audit your browser extensions now - Make a quick inventory of extensions used across your team, especially VPNs, ad blockers, and AI tools, and remove anything you do not absolutely need.

• Treat AI chats as sensitive data - Assume AI conversations can contain confidential or personal information, and set simple rules for your team about what should never be pasted into an AI chat.

• Prefer native apps and official integrations - Where possible, use official web apps or desktop apps from AI providers and trusted VPN clients, not random third party “AI enhancer” or “free VPN” extensions.

• Lock down your standard browser build - For small teams, one practical move is to define a “clean” browser setup with a short approved list of extensions, and block or discourage everything else.

• Train people to spot red flags - Teach staff that high ratings and shiny badges are not enough, and to be especially cautious of extensions that promise free VPN, AI protection, or productivity magic in one click.

🔚 The Bottom Line

This story is a wake up call that your browser is now one of the most sensitive parts of your AI stack. It is not just what you type into ChatGPT or Claude that matters, it is which extensions are quietly sitting in the background watching.

The good news is that a simple extension cleanup and some basic habits can dramatically lower your risk.

💬 Your Take

Are you comfortable with how many extensions you have installed right now, or is this the nudge you needed to do a browser “spring clean” and tighten things up for yourself and your team?

2 comments