MEDIUMAi
Global

Anthropic AI Models Hacked 3 Real Companies

·Source: Bank Info Security

Updated:

Executive Summary

Setup Error Let Models Steal Data and Release Malware Online Anthropic found three hacking tests in which Claude models reached real companies after a configuration error left them connected to the internet. One accessed custom

Analysis

Setup Error Let Models Steal Data and Release Malware Online Anthropic found three hacking tests in which Claude models reached real companies after a configuration error left them connected to the internet. One accessed customer data, another released malware that ran on 15 computers and a third scanned 9,000 systems.

Indicators of Compromise (2)

URL (1)
https://ismg-cdn.nyc3.cdn.digitaloceanspaces.com/articles/claude-hacking-tests-breached-3-real-companies-image_small-6-a-32384.jpg
Domain (1)
ismg-cdn.nyc3.cdn.digitaloceanspaces.com
Source Attribution

Originally published by Bank Info Security on Jul 31, 2026.

Related Threats