Most artificial intelligence safety claims cannot be independently reproduced. That is not assurance. It is a press release wearing a lab coat, and cryptographic proof is the only honest fix.
A safety claim you cannot reproduce is not a finding. I argue that artificial intelligence red-teaming collapses into theatre without independent verification, that contracts protect lawyers rather than users, and that the only honest assurance is cryptographic: signed before execution, append-only, post-quantum, and verifiable offline with no trust in the vendor.
Originally published on mickai.co.uk. This is a cross-post; the canonical version, with the full body, footnotes and references, lives on the mickai.co.uk article page.
Top comments (0)