Hack The Box has launched AI Range Enterprise Edition, a controlled environment for measuring whether AI security agents can perform assigned cybersecurity roles under realistic conditions.

The platform produces role-based scores, pass-or-fail results and performance histories so organizations can retest agents as models, tools and operating conditions change. Hack The Box also introduced AI-augmented penetration tester and SOC analyst roles that measure how well practitioners direct, verify and challenge agent output.

That distinction matters because benchmark performance does not automatically translate into dependable work inside an enterprise. Security agents can drift when their prompts, data, permissions or underlying models change. Repeated role-based testing gives leaders evidence about where automation is reliable and where human intervention remains necessary.

The company first introduced AI Range in December 2025. The enterprise edition turns the concept into a recurring appraisal process, reflecting a broader shift from evaluating models in isolation to testing the complete human-agent system.