Mobile AI agents get safer auto-evaluation
A new framework evaluates AI phone agents step-by-step, catching both task failures and unsafe actions automatically.
A new framework evaluates AI phone agents step-by-step, catching both task failures and unsafe actions automatically.