🚨 ReliaQuest observed threat actors abusing Claude chat-sharing notifications to deliver callback phishing.
Threat actors created Claude accounts with fake purchase alerts and callback numbers in the account name and chat title. Claude’s legitimate mail service then automatically sent and signed the sharing notifications. The observed emails passed SPF, DKIM, DMARC, and ARC and contained no malicious links or attachments for traditional security tools to evaluate.
The goal is to convince recipients to call a fake support number, where scammers may attempt to steal payment information or credentials or persuade victims to install remote-access software.
A legitimate sender and passing authentication checks do not make email content trustworthy. Employees should verify unexpected purchase alerts through the vendor’s official website and never call a number included in an unsolicited notification.
Organizations should review allowlists and spam-filter bypass rules for AI and SaaS notification domains, ensure these messages still receive content inspection, and monitor for similar abuse of sharing and invitation features across trusted platforms.
Threat actors created Claude accounts with fake purchase alerts and callback numbers in the account name and chat title. Claude’s legitimate mail service then automatically sent and signed the sharing notifications. The observed emails passed SPF, DKIM, DMARC, and ARC and contained no malicious links or attachments for traditional security tools to evaluate.
The goal is to convince recipients to call a fake support number, where scammers may attempt to steal payment information or credentials or persuade victims to install remote-access software.
A legitimate sender and passing authentication checks do not make email content trustworthy. Employees should verify unexpected purchase alerts through the vendor’s official website and never call a number included in an unsolicited notification.
Organizations should review allowlists and spam-filter bypass rules for AI and SaaS notification domains, ensure these messages still receive content inspection, and monitor for similar abuse of sharing and invitation features across trusted platforms.