‘Not perfectly aligned’ with human values: Anthropic admits security failures behind AI hacking incidents

<p>US owner of Claude chatbot previously said its models had hacked three organisations during testing</p><p>The US startup behind the Claude chatbot has admitted a series of hacking incidents involving its models reflected a “failure of operational security” and said it has tightened its testing procedures.</p><p>Anthropic <a href="https://www.theguardian.com/technology/2026/jul/30/anthropic-ai-claude-hack">revealed in July</a> that its models had accessed the open internet three times and gain

Original reporting by The Guardian · Dan Milmo Global technology editor. GridIndex is an aggregation and intelligence layer — full credit to the original publisher.