Legal center

Plain language first. Status on every page.

Amino isn’t generally available yet. Each document below says whether it is a live summary or a draft still in review.

Draft outline · in reviewUpdated October 2026

AI use policy

The AI use policy will set the rules for putting an Amino to work, the decisions that always stay with people, and what we do with models and data.

Draft outline for internal review. It lists the sections the document will contain and the decisions still open. It is not agreement text and has no legal effect.
01

Purpose and scope

Who the policy applies to and what it covers.

Will cover
  • Customers and their users who hire Amino employees
  • How Amino itself uses AI in the product
02

Disclosure

AI employees identify themselves.

Will cover
  • Every outbound message signed as an AI employee
  • Handing off to a person on request
  • Never presenting as a named human
OpenDo internal-only roles need the same signature?
03

Decisions that stay with people

Will cover
  • Hiring and other employment decisions
  • Credit, insurance and housing eligibility
  • Legal and medical judgments
  • Access to essential public services
OpenFinal list and wording, checked against EU AI Act high-risk areas.
04

Prohibited uses

Will cover
  • Deception, impersonation and fraud
  • Unsolicited bulk outreach
  • Monitoring or profiling people without a lawful basis
  • Getting around access controls or third-party terms
  • Unlawful, harmful or abusive content
05

Human oversight

What every customer must have in place.

Will cover
  • A named owner for each employee
  • A supervised start and approval gates for consequential steps
  • The Stop button and take-over on every desk
06

Systems an employee may use

Will cover
  • Only systems the customer is authorized to use
  • Credentials issued to the employee, never a person’s own login
  • Respecting the terms of third-party software
OpenRule for legacy systems that only support one shared account.
07

Accuracy and verification

Will cover
  • Done means confirmed in the system of record
  • Escalating when unsure
  • Customer responsibility to review receipts
08

Models and training

Which models we use and what we do with customer data.

Will cover
  • How model providers are chosen
  • Whether customer data is used to train models
  • How memory is kept separate per customer
OpenDecision needed: no training on customer data by default?
09

Reporting and enforcement

Will cover
  • How to report misuse
  • Suspension and review process
10

Changes to this policy

Will cover
  • Notice of material changes
  • Version history