AI Policy & AI Transparency Notice

This notice explains, in general terms, where AI is used in the Platform, what it does, what it cannot do, how it is governed, and what is expected of the humans who use it. It supplements the Terms of Use and the Privacy Policy.

Every AI-assisted output is presented as a suggestion to a human user. AI features do not execute safety-relevant actions autonomously, do not sign records, do not certify work, and do not release aircraft or components. Where an AI suggestion leads to a change in a record, the change is attributed to the human user who accepted it, and that user remains responsible for it.

Authorised personnel must validate AI output against approved data and the organisation's own procedures before relying on it.

• output may be incomplete, outdated, imprecise or factually wrong;
• output may be plausible in style yet incorrect in substance, including fabricated references;
• output may misinterpret abbreviations, technical language, units, effectivity or configuration context;
• output may reflect gaps, errors or inconsistencies in the underlying records;
• output may not reflect the latest revision of an external document or regulation;
• identical questions may produce different phrasing or emphasis on different occasions.

AI output must not be used as approved maintenance data, as an engineering instruction, as the basis for a certification or release decision, as evidence of regulatory compliance, as a substitute for a required inspection or independent check, or as a replacement for the judgement of licensed and authorised personnel.

When a user invokes an AI feature, the request and the context needed to answer it are sent to a model provider acting as subprocessor, solely to generate that response. Customer Data is not used to train generative or foundation models for our purposes or for other customers, and providers are contractually prohibited from training on it. Requests are scoped to the requesting Organisation so that AI features do not surface another Organisation's data.

• AI features are enabled deliberately and are described in the Documentation;
• model providers and configurations are reviewed before adoption and on material change;
• AI interactions are logged so that output can be traced to a request and a user;
• issues reported by users are reviewed and may lead to prompt, retrieval or configuration changes, or to a feature being restricted or withdrawn;
• changes that materially affect AI behaviour in safety-relevant contexts are communicated.

AI features are provided as general-purpose assistance for information handling. They are not qualified, certified or approved tools for airworthiness determination, and their use does not transfer any organisational responsibility to us. Organisations remain responsible for assessing whether and how AI assistance may be used within their approved procedures.

Incorrect, misleading or unsafe AI output should be reported through the legal requests page so it can be investigated. Reports that indicate a potential safety impact are prioritised.