Montana Empire

The attack chain is simple: probe models for invented domains that appear repeatedly, register the most useful names, place phishing or malicious content behind them, and wait for a person (or, increasingly, an autonomous agent) to follow the recommendation.

Since AI models frequently hallucinate phony information, they sometimes point users to websites that don’t exist. Threat actors are now registering these AI-hallucinated domains... “Armed with a prioritized inventory of phantom domains, adversaries proceed to preemptively register those most valuable for attacks.”

If an AI agent executes scripts, pulls data, or downloads software dependencies from an LLM-generated URL without human validation, it can in fact inject malware directly into production pipelines.

Threat actors are using a new technique called “phantom squatting” to trick AI tools into directing users to phishing sites... Threat actors are now registering these AI-hallucinated domains and using them to host phishing sites.

В другом случае исследователи заметили «выдуманный» домен почтовой службы за 51 день до его регистрации... Ресурс распространял вредоносное приложение для Android.

an adversary registered the domain and deployed a weaponized phishing kit impersonating a legitimate marketplace.

The attacker constructed a sophisticated phishing kit named “Montana Empire”... with the ability to route stolen credentials from incoming users to a Telegram-based command-and-control server.

Операторы малвари могли вручную подтверждать одноразовые коды, введенные жертвами, через бота в Telegram.

The attacker constructed a sophisticated phishing kit named “Montana Empire”... with the ability to route stolen credentials from incoming users to a Telegram-based command-and-control server.

Harvesting of national identity document data alongside payment credentials

Операторы малвари могли вручную подтверждать одноразовые коды, введенные жертвами, через бота в Telegram.

The tool automatically scraped the legitimate e-commerce storefront and built a PHP backend with the ability to route stolen credentials from incoming users to a Telegram-based command-and-control server.

Telemetry from the session logs confirmed the adversary leveraged the assistant to scrape legitimate storefronts, engineer the PHP backend and develop a Telegram-based C2 interface for real-time credential exfiltration.