Hackers Forced ChatGPT, Grok, and Google to Help Them Install Malware

Hackers have learned to use popular chatbots as attack tools. They create dangerous prompts that appear among Google's search results and prompt users to run commands themselves, opening up access to their devices.

What is known

According to the cybersecurity company Huntress, attackers engage in a conversation with ChatGPT or Grok on topics of popular search queries such as "how to free up space on a Mac." During the process, they insert a malicious command, then make the chat public and promote it through Google ads. When a regular user stumbles upon this link, they see seemingly useful advice and execute the command, launching dangerous code.

This is how the distribution of AMOS, a Mac trojan capable of stealing data, was discovered. One user searched for how to clean a drive, stumbled upon a ChatGPT link, copied a command into the terminal, and handed control of the device to hackers.

Huntress specialists note that this type of attack bypasses classic signs of danger. There's no need to download suspicious files, open malicious websites, or run programs.

Source: Huntress