AI Transparency Statement

Preferium AI Edge is a system from Preferium AS for technical search optimization, content assistance and visibility measurement. Preferium may be the provider of the AI system. The Customer, the Agency and the End Customer may have roles as deployer, distributor or, in some cases, provider depending on actual use, rebranding and modifications.

The agreed use is technical search optimization, content assistance and visibility measurement. This description is not a regulator classification and not a general confirmation that every customer configuration falls outside the rules on high-risk use. Roles and duties are assessed on the basis of actual function, purpose, data, market, rebranding and substantial modifications. Establishment in Norway does not exclude duties for activity that falls within the scope of the EU rules.

The system can read public website content and supported data sources, generate proposals and, within the selected control level, publish HTML changes at the edge. This may cover titles, meta descriptions, headings, structured data, alternative text and links. In supported deployment flows, published changes are re-checked on the live page and can be rolled back; not every configuration includes every check.

Models and measurement services depend on the confirmed configuration of the specific service. Generating content and measuring visibility or citation are different processing activities and may use different providers. The Subprocessor Register for the service must show who is used and which data is processed. The legal and administrative platform on preferium.com does not itself run a customer's edge generation or citation measurement.

For supported deployment flows the Service may use guardrails, quality checks, logging, control levels, verification and rollback. Not every control applies to every function or configuration, and none of them guarantees that every error is detected.

The Customer must select human oversight according to risk, check regulated and fact-sensitive claims, and be able to stop or roll back publication. Preferium may reject or roll back unlawful, harmful, misleading or platform-breaching output.

Unlawful use and practices prohibited under applicable AI legislation are not permitted and cannot be approved by Preferium. Other regulated or particularly risky applications outside the ordinary SEO service require, where the use is not prohibited by law, written prior approval and a documented legal and technical assessment. This includes the limited fields of use listed in the Acceptable Use Policy.

No customer or reseller may substantially modify the system, change its intended purpose or market it under another designation in a way that creates new provider obligations unless responsibility and documentation have first been agreed in writing.

Information about AI use and the marking of output must be adapted to the specific role, function and mode of publication. Direct interaction with an AI system, machine-readable marking, deepfakes and text on matters of public interest may carry different requirements and exceptions. Human editorial control is not a general exemption from every transparency duty. The Customer must ensure relevant AI literacy, check fact-sensitive claims and comply with the duties that apply to its use.

The Customer must ensure sufficient AI literacy among the employees and contractors who use the Service. Preferium provides the information reasonably necessary for responsible use and investigates complaints about errors, misleading output or breaches of the agreed guardrails.

The EU rules have their own dates of application and scope; national implementation in Norway was still in progress when this version was written. Applicable data-protection, marketing, criminal and copyright rules apply regardless of that legislative work. See the European Commission's AI Act overview.

Preferium does not use Customer Data or visitors' personal data to train general models for other customers. The terms and settings of an underlying AI provider must be assessed and documented for the specific service. This statement does not confirm that an unverified provider configuration has a particular retention period, training opt-out or processing location.

The table shows the default configuration of Preferium AI Edge as read in the product's source code on 19 September 2026. Model versions change; the provider and the kind of data sent are what this table records. A specific Customer's configuration may differ, and the register version tied to its agreement governs.

AI and measurement data are not necessarily processed only in the EU/EEA. Cloudflare may process globally, and some providers are in the United States. Details and transfer mechanisms are stated in the Privacy Notice, the Data Processing Agreement and the Subprocessor Register.

AI output may be wrong, incomplete, non-unique or not protected by copyright. The Customer receives the rights Preferium can grant under the Service Terms, but no guarantee of exclusivity, registrability or that output does not resemble third-party material. The Customer has final editorial responsibility for publication.

We do not guarantee ranking, indexing, AI citation, traffic or revenue, nor the accuracy, availability or continued existence of any third-party model or AI service; providers may change, restrict or withdraw models without notice. The warranty disclaimers and limitations of liability in the Service Terms apply to AI output and to this statement. Errors can be reported to info@preferium.com with the domain, URL, change and requested action.