Claims and boundaries.
What we claim about the platform, what we do not, and where the line sits.
Read the claimsEvidence
What we publish so you can check what we say.
Most vendors ask you to take their claims on trust until a security questionnaire forces the question. We would rather put the checkable material where you can read it first. Every no, zero, and never on this site is scoped to detected values, and the pages below are where that scoping gets tested.
Detection baseline.
A local measurement of PHI detection on a labeled corpus, with the method and the limits alongside the numbers.
Read the baselineHL7 FHIR AI Challenge.
What we built for the challenge, what it proved, and two things we asked HL7 to improve.
Read the writeupSecurity posture.
What is in place, what is being formalized, and how to reach us about a vulnerability.
Read the security page