01AI interactions and outputs
Orion, EVA and enabled assistants use AI to generate or interpret content and propose workflows. Their outputs can be wrong, incomplete, biased or non-unique. They are not professional advice. Users must receive a clear indication of an AI interaction when required; published deepfakes and qualifying public-interest content require the applicable disclosures. Merely linking this policy does not replace a timely in-interface disclosure.
02Context, providers and training
Only authorised context needed for the request may be sent to the enabled provider. Credentials must remain outside model prompts and browser bundles. An adapter or model catalogue entry is not proof of an active vendor contract. General-model training on customer data is not authorised by the standard agreement. Provider logging, abuse monitoring, retention and transfers must be addressed in the actual service schedule; setting store:false is not a zero-retention guarantee.
03Human supervision and effects
An AI proposal cannot approve itself. Authorised effects must bind the concrete action and parameters to the actor, tenant, resource version and approval expiry. The owning product rechecks permission, consent/contactability, price or stock immediately before the effect. Retries must reconcile uncertain outcomes. Normal operations must remain usable when AI is unavailable; private mode requires explicit permission for an external provider.
04Restricted uses and responsibility
Do not use these services for prohibited AI practices, covert manipulation, discriminatory or unlawful surveillance, or solely automated significant decisions without the required legal basis and safeguards. Recruitment, credit, biometric or other regulated/high-risk use is not authorised by a generic software subscription. Huberway and the customer remain responsible for the obligations applicable to their actual provider/deployer roles; no blanket AI Act certification is claimed.
05Output rights and reports
Output rights follow the accepted terms and underlying third-party rights. No output is guaranteed exclusive or copyrightable. Review claims, sources and intellectual-property rights before publication. Report unsafe output, suspected data exposure or unauthorised effects to hello@huberway.com with a minimal description and operation reference, without credentials or unnecessary personal information.