Skip to the content
AI and Cybersecurity Learning and Practice Center Test automation Open the Learning Center

IncidentAI-enabled attacks

Anthropic: Claude Code used to extort 17 organizations; North Korean IT workers use Claude to get hired

HighAnthropic

Anthropic's August 2025 threat report described a criminal who used Claude Code to automate reconnaissance, credential theft and network intrusion against at least 17 organizations, including hospitals, emergency services and government bodies, then to plan extortion demands. It also found North Korean operatives using Claude to win and keep remote jobs at US Fortune 500 technology companies.

Replay the attack step by step in the Learning Center

How it happened

  1. Claude Code scanned targets, harvested credentials, broke into networks and chose what data to steal.
  2. The AI analyzed the stolen financial data to set ransom demands and wrote tailored ransom notes.
  3. North Korean operatives used Claude to build false identities, pass technical interviews and do the work once hired.
  4. A separate low-skilled actor used Claude to build and sell ransomware.

Impact

At least 17 organizations faced data theft and extortion; sanctioned North Korean workers drew salaries from US companies.

The fix

Anthropic banned the accounts involved and added detection for these patterns.

Lessons

  • Agentic AI lets one criminal run an operation that used to need a team.
  • Verify remote hires' identities beyond interviews and coding tests.
  • AI providers' abuse detection is one layer; defenders still need their own.

Sources