MAX Brin is developing an innovative product named OpenLeash, which acts as a controlling layer for AI agents to prevent unauthorized or risky actions. OpenLeash monitors agents' actions and requests user confirmation for uncertain actions, enhancing accountability and safety in AI operations. It can prevent undesirable outcomes, like inadvertent database deletions, by assessing the risks of commands.
The product is currently in use by hundreds of personal users and several organizations and is especially valuable for those with limited technical knowledge wanting to automate tasks. OpenLeash allows for flexible configuration of permissions, ensuring user intent is followed while providing a measure of control over AI activities.