PTAAI-GS v1.0 · The Provable Trust Architecture for AI
Nobody can prove their AI is trustworthy. Until now.
Declared. Audited. Certified. None of it is proof.
{{x}}
Declared
“We have a policy.”
{{wave}}
Audited
“An auditor agreed.”
{{seal}}
Proven
“Check it yourself.”
Most AI governance today ends at the second door. We call it Asserted, not proven.
12requirements
64testable controls
5grades of evidence
4levels of AI autonomy
1verdict nobody else gives
What they are, and how they're tested, is shared under NDA.
The Provable AI Field Guide · preamble to PTAAI-GS
Twenty questions. Can you answer five?
Tick the ones you can prove. The other fifteen are in the free Field Guide.
Open to every sector
Built for anyone who builds, buys or runs AI.
Behind the door
See the full Standard.
Released under our mutual NDA and an evaluation licence. Pilot assessment places are limited.