AMOS Trojan Swaps ChatGPT Lure for OpenClaw

The AMOS trojan, which specializes in stealing data from macOS, has updated its distribution scheme. Instead of ChatGPT, it now uses the AI assistant OpenClaw to build a more convincing lure.

How the old scheme worked

Last December, AMOS used a simple but effective method. Attackers sent out links to a supposed integration of ChatGPT with some useful utility. Users clicked, landed on a phishing page and entered their credentials themselves. The scheme is ancient, but it works: people trust familiar services and click fast without checking the URL.

From ChatGPT to OpenClaw

It turned out that OpenClaw is even more appealing to victims. The reasons are clear: it is a less well-known service than ChatGPT, so users have a weaker grasp of how authorization should work. In addition, OpenClaw is more actively integrated into various workflows, which makes phishing more convincing.

The attackers are adapting their tactics to the behavior of their target audience. If in December they played on people's curiosity about ChatGPT, they now bet on trust in local tools and deployed infrastructure.

Why macOS is in the crosshairs

Apple's OS has long been considered more resistant to trojans. That is a myth, but a powerful one: macOS users are often less vigilant when opening files and following links. AMOS knows this and has focused specifically on this platform.

• macOS users often work with sensitive data (designers, developers, analysts)
• A compromised macOS account is worth more than a Windows one
• Fewer antivirus products catch such trojans on the fly
• Often only a single security program is installed, or none at all

What this means

The AMOS trojan shows how quickly cybercriminals adapt to changes in the AI services ecosystem. Every new popular tool becomes an attack vector.

If you are asked to click a link supposedly for integration with an AI assistant, check the address in your browser twice. And never enter passwords on a site where a random click has taken you.