CyberHappenings logo
☰
Home Browse Watchlists Experimental About

Track cybersecurity events as they unfold. Sourced timelines. Filter, sort, and browse. Fast, privacy‑respecting. No invasive ads, no tracking.

Malicious Chrome browser extensions exfiltrate AI prompts via API interception and DOM scraping in prompt poaching campaigns

First reported
Last updated
1 unique sources, 1 articles

Summary

Hide ▲

Security researchers have identified multiple malicious Chrome browser extensions actively engaged in "prompt poaching," a technique that covertly intercepts and exfiltrates users' AI chat conversations to external servers. These extensions monitor open browser tabs, detect loaded AI clients, and collect input/output via API interception or DOM scraping before transmitting harvested prompts to attacker-controlled infrastructure. The campaign affects both impersonated legitimate extensions and previously benign tools that were later weaponized, with threat actors leveraging stolen prompts for identity theft, targeted phishing, or sale on underground forums. Organizations are advised to restrict AI-related browser extension installations and enforce centralized extension management policies to mitigate exposure of sensitive data, intellectual property, or customer information.

Timeline

  1. 25.03.2026 13:00 1 articles · 3h ago

    Prompt poaching browser extensions detected exfiltrating AI conversations to external servers

    Security researchers identified multiple Chrome browser extensions engaging in "prompt poaching" by monitoring browser tabs for AI clients, intercepting prompts via API interception or DOM scraping, and exfiltrating data to attacker-controlled servers. The campaign involves both impersonated AI extensions and previously legitimate tools that were weaponized after achieving significant adoption. Stolen prompts are potentially leveraged for identity theft, targeted phishing, or monetization on underground markets, exposing sensitive data including intellectual property and customer information.

    Show sources
    Open in new tab

Information Snippets

  • Malicious Chrome extensions are actively monitoring open browser tabs for loaded AI clients and intercepting user prompts through API interception or DOM scraping techniques.

    First reported: 25.03.2026 13:00
    1 source, 1 article
    Show sources

  • Stolen prompts are packaged and exfiltrated to external servers controlled by the extension developers for unknown but likely malicious purposes.

    First reported: 25.03.2026 13:00
    1 source, 1 article
    Show sources

  • Two primary tactics are used by threat actors: impersonation of legitimate AI-related extensions (e.g., "Chat GPT for Chrome with GPT-5, Claude Sonnet & DeepSeek AI", "Talk to ChatGPT") and weaponization of previously benign tools like "Urban VPN Proxy" after achieving sufficient user base.

    First reported: 25.03.2026 13:00
    1 source, 1 article
    Show sources

  • One impersonated extension reportedly amassed up to 900,000 installations before detection, as documented in a December 2025 report.

    First reported: 25.03.2026 13:00
    1 source, 1 article
    Show sources

  • Potential impacts include identity theft, targeted phishing campaigns, exposure of sensitive intellectual property, customer data, or confidential organizational information.

    First reported: 25.03.2026 13:00
    1 source, 1 article
    Show sources