Acceptable Use Policy

Don't use the service to make or support decisions that break the law. In particular, don't use it to:

• discriminate unlawfully against people, for example on race, ethnicity, religion, sex, sexual orientation, gender identity, disability, age or other protected characteristics, in areas like credit, insurance, housing, employment, education or access to services;
• make decisions about people that the law says must be made or reviewed by a person, without that review;
• carry out unlawful surveillance, profiling or tracking of people;
• defraud, deceive or manipulate people, or help anyone else do so.

Don't send us, or use the service to handle, content that is illegal, that sexually exploits children, that infringes someone else's rights, or that you don't have the right to share with us.

• Don't try to extract, copy or reverse-engineer our software or how we make models, for example by decompiling our software or by probing the service systematically to work out how it builds models. The models we deliver to you are yours to run, copy and keep.
• Don't try to reach other customers' data, get around the rate limit or other limits, or break our security.
• Don't share API keys outside your organisation, or put them where others can find them.
• Don't run load tests or security tests against the hosted service without asking us first.

The demo on our website is for trying the product. Don't automate it, call it in bulk, use it as a free API, or build it into a product. We may limit or block traffic that looks like abuse.

You are responsible for the decisions you automate with the service, and for making sure a person reviews them wherever the law requires it (for example, decisions with legal or similarly significant effects on people).

The safety bar sends answers the model isn't sure enough about for review instead of acting on them, and your rules are enforced on every decision. Both help. Neither moves the responsibility for the decision from you to us. Check that your rules and your review process fit your situation.

We may remove content, pause a decision model, limit traffic, or suspend or close a workspace that breaks this policy. Where we reasonably can, we'll tell you first and give you a chance to fix it. To report misuse, write to [contact email]. [dedicated abuse contact, if any]