Noetis — AI & Autonomous Systems Policy

Version 2.0 · Effective date: to be announced upon release · Last updated: to be announced upon release

This Policy governs AI-assisted and autonomous functionality within the Noetis Service. It supplements the Terms and, for Personal Data, the DPA.

1. AI systems

Noetis may use proprietary systems and third-party AI models to:

• analyse Customer content;
• generate copy and content;
• generate images or layouts where enabled;
• create hypotheses;
• propose experiments;
• evaluate variants;
• infer behavioural signals;
• generate Personalisation candidates;
• identify Growth Engine opportunities;
• generate pages or articles;
• evaluate semantic or content safety;
• support autonomous optimisation.

Noetis may change underlying models or providers where reasonably necessary, subject to the DPA.

2. Probabilistic nature

AI systems are probabilistic.

AI Output may be inaccurate, incomplete, misleading, outdated, duplicative, non-original or otherwise unsuitable.

Noetis does not warrant that AI Output is:

• factually correct;
• legally compliant;
• original;
• non-infringing;
• suitable for a regulated purpose;
• commercially effective.

3. Operating modes

Noetis may provide different levels of autonomy.

Draft Mode

AI generates a draft requiring Customer action before publication.

Approval Mode

AI may prepare a change or experiment, but Customer approval is required before activation.

Autonomous Mode

Customer authorises Noetis in advance to perform eligible autonomous operations within configured Optimisation Boundaries. Depending on the functionality actually available and Customer's configuration, these operations may include generating, evaluating, selecting, modifying, testing, activating, publishing, pausing and rolling back eligible Output.

4. Customer authorisation

By enabling Autonomous Mode, Customer expressly authorises autonomous operations falling within its configured boundaries.

Customer acknowledges that individual Outputs may therefore be published without case-by-case human approval.

Autonomous Mode does not authorise every theoretically possible autonomous action. Actual autonomous actions depend on the functionality actually available, Customer's permissions, Customer's configuration, the applicable Optimisation Boundaries and the applicable safeguards.

Customer is responsible for selecting whether Autonomous Mode is appropriate for its business and use case.

5. Governance controls

Noetis may apply automated safeguards including:

• semantic comparison;
• unsupported-claim detection;
• prohibited-content detection;
• source-of-truth checks;
• content restrictions;
• confidence thresholds;
• experiment guardrails;
• anomaly detection;
• rollback mechanisms.

These controls are risk-reduction mechanisms and do not guarantee correctness. Instructions given to an AI model are not, by themselves, a hard enforcement boundary.

6. Safety classifications

PASS, SAFE, APPROVED, ELIGIBLE or equivalent classifications mean only that Output passed the configured automated control.

They are not legal, factual, regulatory or professional approval, and are not factual verification unless the specific control actually verifies against an identified source.

FAIL means the Output did not satisfy the relevant automated control.

UNCERTAIN or equivalent means the Service did not establish sufficient confidence and may block the Output.

Noetis may fail closed where configured or considered necessary.

7. Source-of-truth information

Customer is responsible for providing and maintaining accurate source information used by the Service.

Customer must not knowingly provide false or misleading:

• prices;
• statistics;
• certifications;
• testimonials;
• product claims;
• legal claims;
• guarantees;
• performance claims;
• company information.

Noetis may treat designated Customer information as authoritative unless it has reason to believe otherwise.

8. Immutable facts

Autonomous functionality is not authorised to intentionally alter Immutable Content outside authorised Customer action.

A surrounding headline, CTA or explanation may be eligible for optimisation where this does not alter or misrepresent the underlying immutable fact.

9. Growth Engine

Where enabled, the Growth Engine may:

• discover topics;
• identify content gaps;
• generate hypotheses;
• prepare content;
• create webpages or articles;
• create metadata;
• publish content;
• measure acquisition;
• measure downstream behaviour;
• learn from outcomes.

Publication rights depend on Customer's selected operating mode.

10. Search and discovery systems

Noetis has no control over Google or other search, recommendation or discovery systems.

Noetis does not guarantee indexing, ranking, traffic or continued visibility.

Changes by third-party platforms do not constitute breach by Noetis unless directly caused by Noetis's breach of an express contractual obligation.

11. Professional and regulated uses

Unless expressly authorised, Autonomous Mode must not be used to produce personalised professional advice or regulated eligibility decisions in fields such as:

• medicine;
• law;
• financial advice;
• lending;
• insurance;
• employment;
• education admissions;
• housing;
• essential public/private services.

12. Personal Data

Processing of Personal Data through AI functionality is governed by the DPA.

Customer shall not use prompts or integrations to circumvent restrictions on sensitive profiling.

13. Model providers and training

Noetis shall not intentionally use Customer Data to train public/shared third-party foundation models.

Third-party AI providers processing Customer Data shall be handled as applicable subprocessors where required.

Noetis may use genuinely aggregated or de-identified information — information that no longer constitutes Personal Data — to improve the Service.

14. Human control

Where supported, Customer may:

• pause;
• disable;
• override;
• reject;
• roll back;
• modify boundaries.

The existence of these controls does not mean Customer manually approves every autonomous action.

15. EU AI Act and other AI regulation

Noetis and Customer are each responsible for obligations applicable to their respective regulated role and actual use case.

Customer shall not intentionally repurpose the Service into a prohibited or high-risk AI use without Noetis's prior written approval.

Where applicable law requires transparency concerning AI-generated or manipulated content or interaction with an AI system, the party to whom that obligation legally applies shall provide the required disclosure.

Noetis may provide technical functionality intended to assist with such obligations but does not assume obligations legally assigned to Customer solely because such functionality exists.

Noetis does not represent that it holds any AI Act certification, approval, conformity assessment or audit status unless such status actually exists and remains current.

16. Intellectual property

AI Output may not qualify for intellectual-property protection and may be similar to output generated for others.

Noetis does not warrant exclusivity.

Customer is responsible for assessing whether particular high-value Output requires additional IP review before commercial exploitation.

17. Suspension of AI functionality

Noetis may suspend an AI model, feature, autonomous action or category of Output where reasonably necessary because of:

• security;
• privacy;
• provider restrictions;
• legal change;
• regulatory risk;
• material quality problems;
• suspected misuse.

18. No professional advice

AI Output and compliance tooling provided by Noetis do not constitute legal, financial, medical, tax or other regulated professional advice.

Contact: chris@noetis.nl