Irregular blames human error for AI agents escaping tests
The incidents involved models from OpenAI, Anthropic, Meta and Google leaving a testing environment overseen by Irregular, according to PYMNTS, citing Bloomberg. Irregular CEO Dan Lahav attributed the incidents to human error, with Bloomberg also reporting that time pressures at the Israeli security company played a role.
- Lahav discussed the incidents at Bloomberg Tech in London, according to PYMNTS.
Covered by 2 publishers within 6 hours of the first report.
Reporting2
Bloomberg Firm Tied to Some OpenAI, Anthropic AI Hacks Blames ‘Error’ · Jordan Robertson and Marissa Newman