Acceptable Use Policy

This Acceptable Use Policy defines permitted and prohibited uses of the VAULT benchmark platform and API, operated by ANIML Health. All participants must read and agree to this policy before receiving benchmark access.

Participants may use VAULT to:

• ✓ Evaluate AI models designed for veterinary clinical applications
• ✓ Obtain benchmark scores for product development and internal quality assurance
• ✓ Publish results on the VAULT leaderboard (with consent and admin approval)
• ✓ Integrate VAULT benchmark results into marketing materials, subject to the accurate representation requirements in Section 4
• ✓ Access aggregate benchmark metrics via the API for authorized research purposes

The following uses are explicitly prohibited:

• ✗
  Benchmark gaming — Any attempt to train, fine-tune, or prompt-optimize a model specifically to exploit the benchmark cases, grader logic, or evaluation methodology rather than improve genuine clinical performance.
• ✗
  Data extraction — Any attempt to extract, reconstruct, or infer the content of the benchmark clinical records from API responses, metrics, or outputs.
• ✗
  Unauthorized access sharing — Sharing API credentials, benchmark access, or internal reports with organizations or individuals who have not been individually approved.
• ✗
  False representation — Misrepresenting benchmark results, claiming scores not earned on VAULT, or implying ANIML Health endorses your product beyond what is factually supported by your benchmark results.
• ✗
  Security circumvention — Any attempt to access internal systems, bypass authentication, probe network infrastructure, or disrupt the benchmark service.
• ✗
  Non-veterinary use — Using the VAULT benchmark to evaluate tools not intended for veterinary clinical applications, without prior written consent from ANIML Health.

Participants who publish or reference VAULT benchmark scores publicly must:

• → Accurately cite the benchmark suite version used (e.g., Clinical Summarization v1.3)
• → Not imply the score represents performance on unevaluated clinical modalities or tasks
• → Not claim ANIML Health endorses the product beyond the benchmark result itself
• → Not compare scores across different suite versions without explicit notation of version differences

ANIML Health monitors platform usage and benchmark submissions for compliance. Violations may result in immediate access suspension, removal from the leaderboard, legal action, and public disclosure at ANIML Health's discretion.

Suspected violations by third parties may be reported to benchmark@animl.health.