Skip to main content
How It WorksSandboxPricing
← All features

ACCOUNTABILITY

Human Accountability Layer

Every agent has a named human owner. Formally accepted. Cryptographically logged. Legally defensible.

How it works

The Human Accountability Layer formally binds every governed agent to named humans. An organization admin assigns an owner — the human responsible for the agent’s actions. The owner receives an invitation, formally accepts accountability, and their acceptance is cryptographically logged. For Pro tier and above, an approver can be added. For Business tier and above, a delegate provides coverage. Every ownership change is logged immutably. Agents without accepted owners are flagged as ungoverned.

Technical details

Owner acceptance logged with timestamp, IP, and user ID. Ownership history is immutable. accountability_status: fully_governed | owner_only | pending | ungoverned. Governance score penalty: -15 for no owner, -5 for pending owner.

Compliance mapping

EU AI Act Article 14 (Human Oversight)ISO 42001 5.3 (Organizational Roles)

Use cases

  • Demonstrating human oversight to regulators
  • Incident response — immediately identify responsible party
  • Board reporting on agent accountability coverage

Related features

Agent CertificateArbitrationProduction Readiness
Start governing your agents →Read the standard →