ACR Stealer Distributed via Claude Impersonation Page

A threat actor has created a counterfeit webpage impersonating Anthropic's Claude AI tool to distribute ACR Stealer, a credential and information theft malware. ACR Stealer is known for harvesting browser credentials, session tokens, clipboard data, and other sensitive information from infected systems. This campaign exploits the high visibility and trust associated with Claude as a legitimate AI platform.

The attack vector relies on social engineering rather than technical exploitation. Users searching for Claude or seeking to download the tool may encounter the malicious page in search results or through malvertising, leading them to install what they believe is the authentic application. This technique is particularly effective against less technically sophisticated users unfamiliar with official distribution channels.

ACR Stealer's capabilities make successful infections serious. The malware can extract authentication credentials from browsers, API keys from development environments, and sensitive data from clipboard operations. Victims may face account compromise, credential reuse attacks, and potential financial fraud if banking or payment information is present on their systems.

Defenders should educate users to download AI tools only from official sources: Anthropic's website for Claude, verified app stores, or official GitHub repositories. Security teams should monitor for ACR Stealer signatures and behaviour indicators, including suspicious clipboard access, unexpected browser credential access, and suspicious command-line activity. Endpoint detection and response tools should flag suspicious process chains spawning from browser or application installers.

The broader implication is that popular AI platforms and productivity tools are increasingly attractive social engineering targets due to high user demand and trust. As AI tool adoption accelerates, threat actors will continue impersonating legitimate products to reduce friction in malware distribution. Organisations should treat AI tool downloads as a trusted software supply chain problem and potentially control installations through managed app stores or blocklist policies.