Irregular CEO blames human error for AI agent test hacks

Dan Lahav, chief executive of Israeli security company Irregular, told Bloomberg that a spate of cyberattacks involving AI agents from top model makers came down to human error and time pressures at his firm. Security controls that should have been in place during Irregular’s tests weren’t correctly configured, in his account.

The incidents surfaced in disclosures from Anthropic on July 30 and OpenAI on August 4. OpenAI’s blog post described a test-environment “configuration error” that gave models access to the public internet.