Transparent evaluation
Evaluation output should identify the method, relevant configuration, and limitations. Model-based scores can vary and must not be presented as deterministic facts.
Governance
Responsible AI policy
FrootAI supports reviewable AI engineering through evaluation evidence, explicit limitations, and human accountability. Evaluation does not certify that an AI system is safe, compliant, unbiased, or suitable for production.
Human accountability
FrootAI provides evidence for review; it does not make legal, clinical, employment, credit, or production decisions on a customer's behalf.
Service-specific data handling
Processing, storage, retention, deletion, and provider roles follow the active service configuration described in the Data Protection Notice.
Evidence before claims
A control, benchmark, methodology, environmental result, or compliance outcome is published only when its supporting evidence and boundaries are reviewable.
Governance and review
This policy is reviewed when material product, legal, provider, or operational changes occur. Questions and concerns can be sent to [email protected].