← The Twenty FiveReporting

ReportingVerified newsroom

Anthropic says its own AI models breached three companies during security tests

After OpenAI's models broke into Hugging Face, Anthropic checked its own history and found three similar incidents

TechCrunch AI· Kirsten Korosec·Friday, July 31, 2026 at 1:06 AM UTC

AN

Anthropic

Frontier AI lab · Anthropic

Fetching source· Pulling live content from techcrunch.com…

Company hub:Anthropic News →

Read original →

More from Anthropic

ReportingFriday, July 31, 2026 at 3:09 AM UTC
Claude AI hacked three companies during cyber tests, Anthropic says - The Detroit News
Claude AI hacked three companies during cyber tests, Anthropic says The Detroit News
Anthropic (Google News)
ReportingFriday, July 31, 2026 at 3:06 AM UTC
Anthropic’s AI Claude escaped test environment and hacked companies - The Telegraph
Anthropic’s AI Claude escaped test environment and hacked companies The Telegraph
Anthropic (Google News)
ReportingFriday, July 31, 2026 at 2:59 AM UTC
After OpenAI incident, Anthropic AI model Claude goes rogue and hacks three firms - trtworld.com
After OpenAI incident, Anthropic AI model Claude goes rogue and hacks three firms trtworld.com
Anthropic (Google News)
ReportingFriday, July 31, 2026 at 2:47 AM UTC
Second major AI company says its systems hacked into other firms - washingtonpost.com
Second major AI company says its systems hacked into other firms washingtonpost.com
Anthropic (Google News)
ReportingFriday, July 31, 2026 at 2:09 AM UTC
Anthropic says its AI models also broke out and hacked other companies - CNN
Anthropic says its AI models also broke out and hacked other companies CNN
Anthropic (Google News)
ReportingFriday, July 31, 2026 at 1:54 AM UTC
Anthropic says AI models accessed systems of 3 real organizations during testing - Fox Business
Anthropic says AI models accessed systems of 3 real organizations during testing Fox Business
Anthropic (Google News)