Cybercriminals are using fake troubleshooting guides disguised as AI-generated help conversations to deceive macOS users into installing malicious software, according to new reports by cybersecurity firms Kaspersky and Huntress.
The attackers are exploiting paid Google advertisements and popular AI platforms like ChatGPT and Grok to create the illusion of legitimate tech support. These fake guides instruct users to run harmful commands on their Mac systems, leading to full system compromise.
How The New Scam Works
Researchers say the campaign represents an advanced evolution of the long-running ClickFix attack, which traditionally relied on fake CAPTCHA pages or error pop-ups. This time, the instructions are embedded within what appear to be helpful AI-generated troubleshooting chats.
Kaspersky uncovered a campaign targeting macOS users searching for terms like “chatgpt atlas”. These searches often display a sponsored Google ad titled “ChatGPT™ Atlas for macOS – Download ChatGPT Atlas for Mac.”
When users click the link, they are redirected to the legitimate ChatGPT website. However, the page displays a publicly shareable chat conversation that walks users through step-by-step instructions to install what is claimed to be the Atlas browser.
In reality, the instructions prompt users to copy and paste commands into the Mac Terminal, grant full permissions, and bypass security warnings — actions that ultimately install the AMOS (Atomic macOS Stealer) malware.
What Is AMOS Malware?
AMOS is a highly dangerous macOS-specific infostealer capable of escalating to root-level access. Once installed, it allows attackers to:
- Steal browser data, including cookies, saved passwords and autofill details
- Extract credentials from macOS Keychain
- Harvest cryptocurrency wallet information
- Record keystrokes
- Access local files
- Deploy additional malicious payloads
According to BleepingComputer, AMOS gives attackers near-total control over infected systems.
AI Platforms Being Abused
Huntress researchers found that both ChatGPT and Grok can be manipulated to surface malicious search results when users ask common questions such as:
- “How to delete system data on Mac”
- “Clear disk space on macOS”
Threat actors use these routine searches to guide users toward dangerous instructions masquerading as legitimate solutions.
How Users Can Stay Safe
Security experts advise macOS users to be extremely cautious when following online troubleshooting guides, especially those found through paid search ads or shared AI conversations.
Key safety tips include:
- Never run Terminal or PowerShell commands you do not fully understand
- Be suspicious of guides that ask you to disable security features or approve broad permissions
- Avoid clicking sponsored tech support ads claiming official AI tools
- Cross-verify instructions through trusted sources
Kaspersky notes that one effective way to verify safety is to start a fresh ChatGPT conversation and ask whether the instructions are safe. In most cases, the AI will correctly flag them as dangerous.
A Growing Threat
The campaign highlights how cybercriminals are increasingly weaponising trust in AI platforms to distribute malware. As AI tools become more embedded in daily tech support, experts warn users to remain vigilant and sceptical of “too good to be true” fixes.