KHALM
Model provenance infrastructure
Verifiable records for AI models
A model's record, checked offline.
Against a trust store you control.
KHALM-VMR is KHALM's implementation of the Verifiable Model Record (VMR), an open standard for recording what an AI model is and who stands behind it. A verifier checks a record with no network, and policy packs grade it rule by rule against rules that cite the laws and standards — never against the law itself.
Verifiable Model Record
What the model is.
Who stands behind it.
A Verifiable Model Record is a signed record kept beside an AI model. It states what the model is, how it was trained and who issued the record — as hashes and signed claims, never the training data or the model itself.
KHALM-VMR makes records for any AI model, from any vendor, whose weights the issuer holds: a model is described by its files, so the framework it was trained with and the format it is stored in do not matter. KHALM's own engine is one profile of that, not the shape of it.
A reference pack is not legal advice. A “compliant” answer means the record passes that pack's rule; it is not a finding that the law or standard is met. KHALM-VMR is not certified or approved by any regulator or standards body.
An open standard.
Published.
The Verifiable Model Record carries no company's brand: KHALM is its author, and KHALM-VMR is KHALM's implementation. Version 0.1 is published: the specification and its schemas under Apache 2.0, its test vectors under CC0, and KHALM's reference tools for making, checking and grading records under Apache 2.0, so any implementation can check itself against the same bytes. Anyone can be the authority over the rules a record is graded against, and checking a record is free for everyone.
The standard's own site: verifiablemodel.org. The specification, the vectors and the tools: github.com/VMR-Verifiable-Model-Record/vmr.
Shipping a model someone else has to trust?
If a buyer, an auditor or a regulator needs more than your word for what your model learned — let's talk.