Infostealers are emerging as the fastest-growing cyberthreat of 2025, targeting all operating systems and regions. Kaspersky's findings reveal a new campaign where attackers lure victims with a bait link that appears to lead to a legitimate user guide for OpenAI's Atlas browser for macOS. They achieve this by placing paid ads on Google that link to the official ChatGPT site but redirect users to a chat conversation that looks like a credible installation guide. This clever tactic exploits the trust associated with the ChatGPT domain. Once users click the ad, they encounter a brief conversation meant to guide them in installing the "Atlas browser." If users follow the posted instructions, they end up executing a command in Terminal that downloads a malicious script from a suspicious server. This script, disguised as an installation process, asks for the system password and, if provided, installs a well-known infostealer called AMOS. This malware can harvest sensitive data, including passwords from various browsers and files from the user’s computer. It also installs a backdoor for remote access, allowing attackers to control the machine. To protect against these types of attacks, users should maintain reliable anti-malware software on all devices and avoid running commands from unsolicited sources. Any requests to execute commands should raise immediate red flags, as they often indicate a ClickFix attack. If users encounter suspicious instructions, they should consult someone knowledgeable or verify the commands with an AI bot to understand their purpose and risks. This approach can help counter the growing threat posed by malicious actors leveraging AI to exploit user curiosity and trust.