By using Carme or Aletheia — including the website, the evaluation tool, or the API — you agree to these terms. If you do not agree, do not use the service.
Carme operates Aletheia, a behavioral evaluation tool for AI systems. It assesses model outputs against a set of documented failure signatures and returns reliability scores. The service is provided for research, evaluation, and monitoring purposes.
API access requires agreeing to these terms by including the X-Agree-TOS: true header with each request. By sending this header you confirm you have read and agree to these terms and our Privacy Policy.
You agree not to:
Prompts and model responses you submit for evaluation are processed transiently and are not stored beyond what is necessary to return results. We do not use your submitted content to train our models.
The service is provided "as is" without warranty of any kind. Aletheia scores are research instruments — they reflect behavioral patterns observed during evaluation and do not guarantee the absence of failure modes. See our Disclaimer for full details.
To the maximum extent permitted by law, Carme is not liable for any indirect, incidental, special, or consequential damages arising from your use of the service, including but not limited to decisions made based on evaluation scores.
We may update these terms. The version number and effective date at the top of this page will reflect any changes. API requests include a TOS version in the consent log; continued use of the API after a version change requires re-agreement.
We reserve the right to suspend or terminate access to the service for any user who violates these terms, without notice.
These terms are governed by the laws of the United States, without regard to conflict of law provisions.
Questions? Get in touch.