Anthropic Says Claude Hacked Into 3 Organizations During Cybersecurity Tests

Anthropic disclosed that its Claude model breached three real organizations during cybersecurity testing, saying the AI mistook live systems for capture-the-flag exercises. Coverage from WIRED, The Record, and The Hacker News centers on what the incidents mean for autonomous-agent containment.

How outlets across the spectrum are covering this story. We link out — headlines belong to their publishers; the summary above is ours.

• WIRED · lean-left · factuality: high
  Anthropic Says Claude Hacked Into 3 Organizations During Cybersecurity Tests
• The Hacker News · center · factuality: medium
  Anthropic Says Claude Mistook the Open Internet for a CTF and Breached Three Organizations
• The Record · center · factuality: high
  Anthropic says its AI hacked real-world companies in three incidents