Legal
Acceptable
use policy
Last updated 21 September 2026. This policy applies to every vendor workspace and to every end user of the embedded assistant.
Fair housing
The assistant must not be used to filter, rank, describe or steer units, applicants or tenants by race, colour, religion, national origin, sex, familial status, disability or any other characteristic protected by applicable law. The assistant refuses such requests, and attempts to work around the refusal are a breach of this policy.
Decisions the assistant does not make
The assistant does not give legal advice, does not make tenant screening decisions and does not run eviction workflows. Vendors must not map actions that would have it do so.
Actions
Every Write action requires the end user’s confirmation and every Sensitive action requires a typed confirmation. Vendors must not build around these controls, disable audit logging or use the assistant to make changes on behalf of users who have not asked for them.
Security and abuse
Do not probe, scan or test the service without written permission, attempt to access another workspace’s data, send malware, or use the service to send unsolicited messages. Do not attempt to extract model weights or to use retrieved content and API responses to inject instructions into the assistant.
Lawful use
Use the service only in compliance with the laws that apply to you, your end users and the data you process.
Enforcement
Breaches can result in suspension of a workspace or of an end user’s access. Vendors are told what happened and why.