Mac users beware: Scammers are hijacking Claude chats and Google ads to spread malware

Criminals abused Claude’s “Shared Chats” feature to place fake installation instructions that led to information-stealing infections.
Fraudulent chats promoted through Google Ads, displaying authentic Claude URLs to trick Mac users
The campaign used ClickFix tactics, spoofed “Apple Support,” and avoided targeting Russian-language systems.

Experts have warned that cybercriminals are abusing legitimate Claude and Google Ads services to trick Mac users into installing data-stealing malware on their devices.

Recently, security researcher Berk Albayrak discovered and revealed a new campaign on LinkedIn, related to a feature called “Shared Claude Chats,” which allows users to create clickable links from previous conversations they have had with the AI. That way, other people can see those specific chat sessions through a public URL.

According to Albayrak, hackers have created conversations in which the platform displays instructions on how to install Claude Code (a command-line coding assistant). However, the instructions are nothing more than the standard ClickFix scam: they tell the user to open Terminal and paste a command, which sets off a chain reaction that results in an information-stealing infection.

Advertising the scam on Google

The thread was created by an account called “Apple Support”, which will likely add to its legitimacy. Those with a keen eye, however, could easily spot the trick, as the chat has a disclaimer at the top, warning that the content below might be unverified or unsafe.

But creating a fraudulent conversation is only half the process: victims must somehow get there.

That’s where Google Ads comes into play. The criminals were able to purchase ads on Google’s ad network, meaning that people searching for “Claude Code on Mac” would receive this chat at the top of the search engine results page. To make matters worse, those who hovered over the link or checked where it leads would see “claude.ai,” Claude’s authentic URL.

Albayrak did not say how many people could have been compromised in this way, but beepcomputer found that the malware does not work on Russian-speaking computers, suggesting that bad actors actively avoid attacking Russians.