Acceptable use policy

Do not use output to make or materially influence an automated decision about an individual, including hiring, promotion, discipline, credit, insurance, housing or access to services.

Material you have no right to disclose; special-category personal data; regulated health or payment card data; malware; or content that is unlawful, harassing, or infringing.

Do not construct a persona intended to represent a specific identifiable living person without their consent, and do not present generated output as that person's actual statements. Generated debate content must be labelled as simulated wherever it is shared outside your organisation.

No credential sharing, scraping, reverse engineering, circumvention of usage limits or access controls, penetration testing without written authorisation, or use of the service to train a competing model.

We investigate reported breaches, may remove offending content, and may suspend access immediately where there is risk of harm or legal exposure. We aim to give notice and an opportunity to remedy where it is safe to do so.