Executive Summary
Researchers at OX Security have identified a large-scale data theft campaign leveraging two rogue Google Chrome extensions that impersonate the legitimate AITOPIA AI sidebar tool. These extensions, which managed to compromise over 900,000 users, monitor browser activity and specifically target interactions with ChatGPT and DeepSeek to exfiltrate proprietary conversations and browsing history.
The attack chain involves the extensions monitoring browser tabs via the `chrome.tabs.onUpdated` API. Once a target URL (e.g., chatgpt.com) is identified, the malware scrapes the DOM for prompts, responses, and session IDs. This data is Base64-encoded and transmitted to command-and-control (C2) servers every 30 minutes.
This threat poses a significant risk to enterprise security as it directly captures sensitive intellectual property, business strategies, and PII shared with LLMs. Furthermore, the exfiltration of full browsing history provides attackers with insights into organizational structures and session tokens, facilitating further targeted attacks.