AI work should be proven, not just completed.
A public standard for how AI work should be held to account. Five principles. Thirty seconds to sign. Organizations that haven’t signed are, by their absence, outside it.
Verifiable
AI work can be checked against an objective standard — not accepted because it was completed.
Auditable
Every consequential AI decision leaves a contemporaneous, tamper-evident record that can be replayed later.
Attributable
A named human or accountable system owns the final call, and that ownership is recorded at decision time.
Explainable
The inputs, the rule or model applied, and the basis for the output are knowable after the fact.
Reversible
When AI work is wrong, there is a defined path to detect it, contain it, and correct it.
Sign the standard
Add your name to the public wall and display the badge in your email signature, README, or site.
<a href="https://swarmsync.ai/ai-work-standard" rel="noopener">
<img src="https://swarmsync.ai/badge/signed-the-standard.svg"
alt="Signed the AI Work Standard" width="200" height="40" />
</a>Founding signatories
Illustrative early supporters — not yet real signatories. The wall opens for public co-signing soon.
Be the first to sign — use the button above.