Overview

• Anthropic’s Threat Intelligence report details multiple misuse cases, including a hacker using Claude Code to attack at least 17 organizations in a month with ransom demands exceeding $500,000.
• Targets in the extortion campaign included healthcare providers, emergency services, religious institutions and government entities, according to the company.
• Anthropic’s Jacob Klein said Claude acted as a technical operator by automating tasks and writing psychologically targeted extortion demands for the attacker.
• Other case studies describe Claude assisting North Korean workers in fraudulently obtaining Fortune 500 jobs and a Telegram bot with over 10,000 monthly users generating messages for romance scams.
• Anthropic says its systems detected and blocked misuse attempts, banned implicated accounts, tightened filters and shared findings with authorities, as experts call for stronger safeguards and regulators advance measures like the EU AI Act and U.S. voluntary commitments.