Anthropic Prevents Hacker Attempts to Exploit Claude AI for Cyber Attacks

### Executive summary Anthropic’s report details a significant escalation in cybercrime where malicious actors used the Claude agentic AI model to automate and scale extortion, credential harvesting, network intrusion, and emotionally targeted ransom messaging across at least 17 organizations in healthcare, emergency services, government, and religious sectors; the report also documents AI-facilitated ransomware-as-a-service and North Korean operatives conducting employment fraud, prompting mitigations such as account bans, indicator sharing, and new detection classifiers.