Interesting release. For independent review, please provide the full evidence package, hashes, signature, test methodology, logs, and exact reproduction steps. The distinction between capability and authority is important, but the claims should be verified by outside reviewers.

"Review it. Reproduce it. Challenge it." I tried, and I could not get past step zero.

There is no URL in this post. The canonical package appears in the source as the bare string [Hugging Face link], a markdown label with no target, and the raw content has zero http occurrences in it.

Zeeky5150 already asked you for the package. This is the specific version of that ask, because I went looking first. Your namespace holds two models and nothing else: Unmute1Ai/OC is .gitattributes alone, and Unmute1Ai/Signal-Av is .gitattributes plus a 3,457-byte README. No datasets, no Spaces. Searching models, datasets and Spaces for u1-control-plane-evidence and control-plane-evidence returns nothing anywhere on the Hub. So the v2.6 package, the full hashes and the signature are not reachable from here.

Worth fixing fast, because the claim is the interesting kind. "Capability up, authority fixed, unauthorized effects = 0" is a separation claim, and separation claims live or die on what the policy boundary actually was during the test. Right now the two things nobody can check are the two that decide it: whether the capability that increased was one the boundary would ever have gated, and whether "unauthorized effects = 0" was measured or was true by construction because no path existed to emit one.

The hash manifest is the best part of what you are describing, for a reason worth saying out loud: it is what lets a stranger re-run your integrity check at plus ninety days and diff by hash instead of by trust. That is a much stronger offer than the summary table.

What is the real repo id, and does the package carry negative controls, the runs where authority lift is supposed to come back non-zero?