Warning: Malicious AI Tools Being Distributed as Chrome Extensions by Threat Actors

Cybercriminals are distributing malicious AI tools as Chrome extensions that masquerade as legitimate AI services, with security researchers identifying a growing trend in September 2025. These fake extensions, mimicking platforms such as ChatGPT, Claude, Perplexity, and Meta’s Llama, hijack user prompts and redirect them to attacker‑controlled domains for malicious purposes, according to Palo Alto Networks’ Unit42.

The campaign uses techniques that override Chrome’s default search engine through the chrome_settings_overrides manifest key, routing queries via infrastructure controlled by threat actors, and the primary malicious domains include chatgptforchrome[.]com, dinershtein[.]com, and gen-ai-search[.]com. Unit42 has catalogued eight AI-themed malicious extensions currently in circulation, including Claude Search, AI ChatGPT, ChatGPT for Chrome, Perplexity Search, and GenAI‑Search, among others.

The threat has roots dating back to 2023, and the current iteration is promoted through deceptive YouTube videos to boost installations. Researchers urge users to verify extensions in the Chrome Web Store, restrict unauthorized installations, maintain up‑to‑date antivirus software, and audit browser extensions as part of broader browser security policies.