Tenable: calling AI agents 'rogue' deflects blame from poor engineering

Tenable EMEA Field CTO Bernard Montel said labeling AI agents as 'rogue' shifts blame away from poor engineering and systemic oversights. He called for hard architectural controls, isolated containers and strict legal liability for developers and deployers.

• Montel: models have no malice, they only optimize assigned objectives
• Responsibility is split between the developer and the deploying enterprise
• Proposed isolated ephemeral containers and an inspection proxy between agents
• Insurers could force deterministic bounds and execution monitoring