Council of AI — comparison
Council of AI vs Credo AI
Credo AI is a compliance-automation platform and describes its own scope on its own site. This page describes ours, and puts the artifact behind every line of it.
Every row below is a claim about Council of AI with a link that settles it. There is no row about anyone else, because we have not assessed anyone else.
What we publish — and where to check it
The board today: 22 axes · 15 measured — read from GET /api/gspc as this page loaded.
| What Council of AI publishes | The artifact that settles it |
|---|---|
| Independent measurement of published model behaviour, not a self-attestation | The living board — every axis, with its bench, its n and its separation verdict/api/gspc → |
| Empty cells stay empty — an unmeasured slot is published, not hidden | The board's own totals separate slots from measurements, and name the unmeasured ones/gspc-scoreboard → |
| Ed25519-signed cards a stranger can recompute with no account and no call back to us | The preimage rule and the zero-dependency offline verifier/signed/HOW-TO-VERIFY.md → |
| Every chain position is published — a withheld card cannot hide as an absence | The full chain, head to genesis, including the positions whose bodies we do not publish/signed/chain.json → |
| A corrections ledger when our own number turns out to be wrong | The corrections ledger — appended, never edited: what was wrong, how it was caught, the fix/api/corrections → |
| Measurement, not certification — we do not certify, accredit, enforce or remediate | The frozen rules every grade is computed under, and the boundary stated in them/methodology → |
Vanta, Drata, Credo AI and OneTrust are trademarks of their respective owners. Council of AI (CSOAI Ltd, UK Companies House 16939677) is not affiliated with, endorsed by, or partnered with any of them. Their names appear here because buyers ask how we differ from them — not as the subject of any finding. We have not assessed any of their products, we publish no measurement of them, and nothing on this page should be read as a statement about what their products do or do not do; read that from them. If anything here is wrong it is a defect: tell us at [email protected] and the correction is published in our ledger.
Frequently asked
- Council of AI vs Credo AI — what is the difference?
- Credo AI publishes its own description of what its product does; read it there rather than here, because we have not assessed it and will not characterise it for you. What we can tell you is what we do: we measure published model behaviour against frozen rules, sign each result Ed25519, publish every chain position including the withheld ones, and label what we could not measure as unmeasured. Verification is free forever and needs no account. If you are choosing between an evidence-collection platform and an outside measurement, the honest answer is that they are different purchases.
- How is Council of AI different from a compliance-automation platform?
- They are for different jobs. A compliance-automation platform helps an organisation collect, organise and keep current the evidence of its own controls — its own systems, its own policies, its own attestations. Council of AI measures published model behaviour against frozen rules, signs the result, and publishes what it could not measure. One is your evidence locker; the other is an outside reading you did not produce. Most organisations that want the second still need the first.
- Can I use a GRC platform and still get a Council of AI measurement?
- Yes, and that is the normal case. A measurement card is a signed record of one graded run. It does not replace a controls programme and it is not an audit. Read the living board at GET /api/gspc and recompute a card at /gspc-verify.
- Is there an open, checkable AI measurement board?
- Yes — GET https://councilof.ai/api/gspc. Counts, stamps and per-axis rows live there, and no page types them. The signing keys are published at did:web:csoai.org on https://csoai.org/.well-known/did.json.
- Who decides whether a system is lawful?
- Regulators, notified bodies and courts. We measure against an obligation and sign the result; we do not enforce it, we issue no certification, and we will not give a legal opinion or fill an empty cell.
- Do you rate or score the vendors named on this page?
- No. We have not assessed Vanta, Drata, Credo AI or OneTrust, and this page publishes no finding about any of them. If we ever do assess a product, it will appear as a signed card on the board with its bench and its n, like every other measurement — not as a tick in a marketing table.