Перейти к содержимому

AI Usage Disclosure

Версия 2026-11-04

Арабский язык является единственной аутентичной версией условий и политик; английский — это перевод для справки. При расхождении в формулировках или значении преобладает арабская версия в пределах, разрешённых законом, без ущерба обязательным правам или исключения ответственности за вводящий в заблуждение перевод.

1. Purpose and scope

Arabic is the authoritative version of this document; English is a reference translation. Arabic prevails in a conflict to the extent permitted by law, without reducing mandatory rights or excluding responsibility for a misleading translation.

This disclosure explains how “Bayan Flow” uses artificial intelligence, what is sent to model providers, what the limits of the Output are and what your responsibilities are when you use it. It is part of your agreement with us together with the Terms of Service, the Privacy Policy and the Acceptable Use Policy.

2. Where AI is used

Uses available by Plan/settings include reply drafting, editing, translation, templates and brand voice, knowledge retrieval/enrichment, gap analysis/question grouping, integrations and bots.

In assistance mode, users review drafts before sending. An integration with automated replies enabled may send without per-message human review. Background knowledge/gap jobs may also run; model use is not limited to immediate staff requests.

3. What is sent, and to whom

3.1 Data necessary for the task may include text/messages and context, template/brand-voice instructions and knowledge excerpts.

3.2 Models may vary or switch automatically by task/availability. Supplier review must include fallbacks and failure routes; hiding model names in the interface does not remove required disclosures.

3.3 Do not enter secrets or unnecessary data; including them in submitted text may cause processing, and automatic masking is not guaranteed to detect everything.

3.4 The Privacy Policy addresses provider roles, international transfers and retention.

4. Content use and licence limits

4.1 Your subscription does not grant Bayan or its providers a general licence to train or fine-tune general models on your Content, Output or documents. Processing remains limited by its lawful purpose, authorised instructions and data-protection duties.

4.2 Provider practices concerning security retention, abuse review and training depend on the applicable service, plan, settings and contract. A particular model name is not a guarantee of zero retention or no human review by its provider.

4.3 Improving the Service using aggregate statistics does not authorise reuse of personal or confidential Content outside its lawful purpose. Removing some identifiers does not guarantee anonymisation.

4.4 Bayan remains responsible for applicable disclosure, provider-selection and processing-control duties; using an external provider does not release it from its obligations.

5. Accuracy and limits

5.1 Language models predict likely text and do not “know” facts. They can be wrong, invent plausible-sounding information, give outdated information, misread context, show bias, or err in translation, numbers, dates and names.

5.2 In knowledge-grounded replies we instruct the model to rely on excerpts of your documents and not to invent facts about your organisation when it finds nothing relevant. This does not guarantee accuracy: a reply is only as good and current as your documents, and retrieval can miss or misjudge a passage.

5.3 Quality varies across languages and fields, and Arabic, with its dialects and local style, may need closer review.

5.4 Output is not legal, medical, financial, tax or professional advice.

6. Human review is your responsibility

6.1 Review drafts, sources, numbers and units before sending, publishing or deciding, especially customer commitments.

6.2 For automated replies, the authorised customer defines scope/knowledge, assesses suitability, supervises operation and provides human escalation. Disable uses unsuitable for their risk; per-reply review before sending is not presumed.

6.3 Do not rely solely on Output for significant employment, credit, insurance, healthcare, rights or essential-service decisions.

6.4 Customer responsibility for Content/settings does not remove Bayan's liability for its breaches or non-excludable duties.

6.5 Check prices, dates, return and warranty terms, any customer promise, and whether the source is current and applies to the case. The Service may miss a knowledge gap or cite a source that does not support the entire answer; no warning is not proof of correctness. Review templates, knowledge and settings when your organisation's policies change or an unsuitable reply is identified.

7. Prohibited and high-risk uses

Using AI in Bayan for any of the following is prohibited; for detail see the Acceptable Use Policy:

• Generating unlawful, harmful, fraudulent, deceptive or impersonating content.

• Circumventing the safeguards of the models or the Service, or trying to extract their internal instructions or the data of others (including prompt injection).

• Mass automated sending of unsolicited messages.

• Fully automated decisions with legal or serious effects on individuals.

• Giving legal, medical or financial advice to others as if it came from a specialist.

We reserve the right to restrict requests that we judge to breach this.

8. Personal data in your requests

8.1 Data roles follow facts/law under the Privacy Policy and processing agreement, not automatic labelling of every user as controller.

8.2 Minimise personal data and establish required lawful bases, notices and authorisation. Sensitive, health/credit and minors' data need applicable assessment and specific agreement; uploading alone does not permit processing.

8.3 Do not enter passwords, keys or full payment details; masking some identifiers does not ensure context cannot identify someone.

9. Being open with your own customers

9.1 If the Law or commercial honesty requires telling your customers that a reply or content was prepared with the help of artificial intelligence, that is on you.

9.2 Do not use the Service to impersonate a particular human, and do not lead your customers to think they are talking to a person if a reply is fully automated.

9.3 If one of your customers asks how their data is processed or objects to automated processing, respond according to your role and statutory duties. We assist where we process data on your behalf under the processing terms, and fulfil our independent duties where we are a controller under the Privacy Policy.

10. Protection and its limits

Bayan is required to take measures appropriate to the data and processing risks, limit permissions, minimise data and disclose AI limitations. Customers are responsible for selecting automation scope and appropriate supervision under section 6.

A source, filter or usage limit does not guarantee correct Output or detection of every error or misuse. These provisions are not a security certification or a guarantee against every incident; each party's statutory and contractual duties remain.

11. Rights in Output

The Terms of Service govern rights in Output and responsibility for using it. Output may be similar across different users, and that gives you no exclusive right to a general idea or phrasing. You are responsible for ensuring that Output you publish does not infringe the rights of others.

12. Reporting a problem

If AI in Bayan produces a harmful or biased output, or one that shows data you should not see, or you notice unexpected behaviour, use the available feedback channel or tell us through the “Contact us” page describing what happened and when, without sending us sensitive content in the report. This does not delay any mandatory legal duty to handle or report incidents.

13. Principles we follow

We are guided in designing and running the Service by the AI ethics principles published by the Saudi Data and Artificial Intelligence Authority: fairness, privacy and security, humanity, social and environmental benefits, reliability and safety, transparency and explainability, and accountability and responsibility. In practice: keeping a person as the decision-maker, minimising the data sent, separating customers' data, disclosing limits and receiving reports.

14. Changes

We update this disclosure whenever the way we use AI, our providers or the Law change, and its version and date appear at the top of the page. For material changes we notify you before they take effect, as the Terms of Service provide.

15. Languages, sources and reliance

Answer, translation and retrieval quality varies with language, subject, model and supplied Content. Supporting a language or showing an example does not guarantee an identical result for every question or customer.

A citation does not mean every sentence is supported by that source, and the absence of a warning does not establish correctness. Review the response and its source before relying on it, and update company knowledge when policies change. The Terms of Service liability provisions apply without reducing mandatory rights.

AI Usage Disclosure · Bayan Flow