5,000 tests per agent
Enterprise AI is getting too smart to control. The Artificial Intelligence Underwriting Company just raised $40M to audit and certify agents with a 5,000-test standard called AIUC-1. Co-founded by an early Anthropic hire and former METR COO. #AI #EnterpriseAI #AISafety #AIAudit
What the video says
Enterprise AI is getting too smart to control. To fix this, the artificial intelligence underwriting company is launching an independent safety standard.
They run agents through 5,000 tests to detect jailbreaks and hallucinations. While AI runs the evaluations, humans verify the final safety reports.
Now they have raised a $40 million Series A to audit and certify Enterprise agents.
