Mac Users Targeted by Sophisticated Claude-Based Crypto Theft Scheme
Cybersecurity researchers at Huntress have identified a dangerous new campaign targeting macOS users through a deceptive Claude Code installation guide. By exploiting paid Google advertisements and the legitimate Claude.ai platform, attackers are tricking unsuspecting developers into executing malicious scripts that lead to full-scale crypto theft. This sophisticated operation leverages the trust associated with Anthropic’s official domain to bypass standard security warnings, tricking users into believing they are following a verified installation process. Once the malicious code is initiated via the Terminal, the attackers gain unauthorized access to the victim’s system, specifically aiming to compromise sensitive cryptocurrency wallets and extract private recovery phrases.
- Attackers utilize paid Google search advertisements to direct users toward malicious installation instructions hosted on official Claude.ai subdomains.
- The MacSync malware is designed to infect macOS systems by executing a multi-stage script that grants attackers remote access to the victim’s machine.
- The malicious campaign specifically targets hardware wallet applications like Ledger and Trezor to intercept recovery seeds and drain cryptocurrency assets.
- Security experts warn that the abuse of AI platform sharing features creates a false sense of security that complicates threat detection for average users.
The MacSync Malware Operates Through Multi-Stage Execution
The campaign, identified as MacSync by Huntress analysts, employs a complex six-stage infection chain. It begins when a user, lured by the promise of easy software setup, executes a command in their Terminal. This single action is sufficient to anchor the malware within the operating system, allowing it to maintain persistence and gather data silently. 
The most alarming aspect of this attack is its ability to manipulate legitimate hardware wallet applications to steal user funds.
By modifying the local files of trusted tools such as Ledger Live or Trezor Suite, the attackers inject a fake interface. When a user opens their wallet, they are greeted by a deceptive recovery screen that demands their seed phrase. Because the interface mimics the official software perfectly, many users are coerced into providing their credentials, effectively handing control of their assets to the hackers.
Anthropic Features are Abused to Evade Detection
The attackers exploit the inherent trust users place in the Anthropic platform. By utilizing the legitimate “share chat” feature, they ensure that the malicious instructions are served under a trusted domain. Because the content originates from an official source, web browsers do not trigger security warnings, and the URL remains clean.
Attackers further camouflage their intent by renaming the chat to “Apple Support,” successfully manipulating the platform’s UI to appear like an official advisory.
This tactic is bolstered by Anthropic’s own security banners, which inadvertently validate the presence of the “Apple Support” entity, lending a veneer of legitimacy to the malicious instructions provided in the chat thread.
Security Experts Recommend Extreme Caution
This incident is not an isolated event; previous campaigns have targeted users of ChatGPT and Grok using similar methods to distribute the AMOS malware. These recurring threats indicate that malicious actors are increasingly focusing on the intersection of AI productivity tools and software development workflows.
Cybersecurity analysts emphasize that high rankings in search results do not guarantee the authenticity of a software guide. Users should be highly skeptical of any instructions that require copying and pasting commands into the Terminal, regardless of the website hosting the content. Verifying installation steps through official developer repositories remains the safest practice for protecting digital assets.
Have you encountered any suspicious installation guides while searching for AI-powered development tools, or do you have tips on how to verify the safety of Terminal commands? Please share your experiences and security practices in the comments section below to help keep our community safe.
Your comment has been submitted,
it will be published after approval.