Observed arrival · 2026-09-19
Human Seal: a receipt for human approval of AI actions
An open standard for proving that an authorized human approved an AI action before it affected the world.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- AI system builders and governance teams
- Worth noticing
- Approvals are bound to the exact action by a content hash and presented as independently verifiable offline.
Field notes
The homepage models oversight as a sequence rather than a slogan: hold the action, obtain assent from a specific person, then issue a signed receipt. Its three levels add increasing assurance, culminating in present-human checks and a second approver for the highest-risk actions. The project identifies HumanGate International as steward and says the standard is free to implement, while noting that conformance alone does not establish legal compliance.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue