---
title: "Responsible AI Statement | Ilmiya"
description: "Our approach to AI in education, its limits, and the role of human judgment."
url: "https://ilmiya.com/responsible-ai-statement/"
language: "en"
---

# Responsible AI Statement

Our approach to AI in education, its limits, and the role of human judgment.

AI can help you prepare a lesson or work through an unfamiliar idea. Teachers and the schools or educators providing the service remain responsible for the educational decisions they make with AI.

This statement explains Ilmiya’s approach to AI. The [Acceptable Usage Policy](https://ilmiya.com/acceptable-usage-policy/) sets the rules for using AI features. The [Global Privacy Policy](https://ilmiya.com/privacy-policy/) describes personal information, and the [Data Protection Addendum](https://ilmiya.com/data-protection-addendum/) sets the contractual requirements for handling customer data. This statement does not replace those documents or change a signed agreement.

<a id="what-this-covers"></a>

<a id="scope"></a>

## Scope

This statement covers features that generate or summarize content, suggest learning activities, help answer questions, or process spoken interactions. Schools, other education organizations, and independent educators can provide these services under an agreement with us. Independent educators entering that agreement must have legal capacity to contract.

Depending on the feature you use, an AI provider may receive your prompt, earlier messages, uploaded material, or audio and transcripts to produce a response.

An AI reply is generated content. A teacher may not see your AI conversation, even if you can also send messages to your school through Ilmiya. Messages to people and AI conversations have different audiences, and a message does not guarantee an immediate response.

Voice and AI services are optional. Learners may decline them while retaining the core learning service. The relevant notice and any required permissions must precede activation.

<a id="our-commitments"></a>

## Our Commitments

-   **Educational purpose.** Ilmiya’s AI features support teaching, learning and educational administration. Educators retain responsibility for the instructional decisions they make.
-   **Customer Data.** The DPA’s restrictions on sale, advertising, unrelated profiling and model training apply to AI features. The detailed training rules below remain in force.
-   **Third-party providers.** A model provider engaged by Ilmiya for Customer Data is subject to the DPA’s subprocessor requirements and appears in the [Subprocessors and Service Providers](https://ilmiya.com/sub-processors-list/) register. Information sent to it must be protected in transit and used only for the authorized service under the applicable processing terms.
-   **Transparency.** AI-powered features and generated content should be identifiable to the people using them. The disclosure and safeguarding requirements below apply to the enabled activity.

<a id="human-judgment-in-education"></a>

<a id="human-oversight"></a>

## Human Oversight

Our aim is to help educators spend more time teaching and to give learners useful support. AI output can contain false statements, invented references or biased assumptions. It can misunderstand a learner’s question or produce an answer that is unsuitable for their age, curriculum or circumstances.

If you teach with AI, check the material before using it in a lesson. Schools and educators must follow the [Acceptable Usage Policy](https://ilmiya.com/acceptable-usage-policy/) when using AI in decisions about a learner. Someone with the authority and relevant knowledge must review consequential decisions, including grades of record, admission or discipline. The reviewer must understand the decision and be able to change it. An AI-generated answer is not evidence that a student has mastered a subject. Human review does not remove applicable AI-specific duties or determine whether a system is legally classed as high risk. Students retain any legal right to information, to challenge a significant automated decision, and to obtain human intervention.

AI is not a qualified professional or an emergency service. A learner who needs help with a safety concern should contact a trusted adult or the appropriate local service. A teacher or institution should provide a route to human support when AI is used in learning.

<a id="prohibited-uses"></a>

## Prohibited Uses

The [Acceptable Usage Policy](https://ilmiya.com/acceptable-usage-policy/) applies to AI features. In particular, users must not:

-   Generate unlawful content, exploit or endanger minors, or use AI to harass or deceive another person.
-   Present generated material as human-authored where that misrepresentation would mislead or harm others.
-   Make a consequential automated decision about a learner without the required meaningful human review or other safeguards required by law.
-   Extract or reconstruct another person’s personal information, or attempt to re-identify protected data from a model or its output.
-   Circumvent the safety or access controls applied to an AI feature. Authorized security testing and lawful reporting remain subject to the applicable agreement and Acceptable Usage Policy.

<a id="institutional-controls"></a>

## Institutional Controls

The Customer determines whether and how AI features are used in its educational offering, within its lawful authority and the purchased Services. Authorized administrators manage available permissions and configuration. Ilmiya must follow the Customer’s lawful documented instructions, including instructions to disable an optional use. The available controls depend on the feature and configuration supplied.

A preview, beta or experimental feature must be identified as such before use. Its availability and support follow the [Support Policy](https://ilmiya.com/support-policy/) and the governing agreement. A preview designation does not remove the DPA’s protections or applicable legal duties.

<a id="accuracy-disclaimer"></a>

## Accuracy Disclaimer

AI output is probabilistic and may be inaccurate, incomplete or unsuitable for a particular learner even when safeguards are used. Ilmiya does not warrant the accuracy of generated content or a particular educational result. Output must be checked before instructional use, and the governing agreement determines the applicable warranties and remedies. These limits do not remove mandatory rights or Ilmiya’s express contractual commitments.

<a id="personal-information-and-ai"></a>

## Personal information and AI

We do not currently train Ilmiya models on customer data, and we do not allow third-party providers to use customer data to train their models. The [Data Protection Addendum](https://ilmiya.com/data-protection-addendum/) contains the contractual restrictions on model training.

A future customer-specific Ilmiya fine-tuning feature, including one using low-rank adaptation (LoRA), would require a separate, informed opt-in under written terms before any eligible data was used. Those terms would need to explain the data, purpose, model access, retention and available choices. A customer’s agreement cannot authorize a use that the law prohibits or replace an individual’s consent where it is required. This statement grants no training permission, and we cannot expand those permissions by changing this page.

Providing context to generate a response, retaining a chat history, and training a model are different activities. Schools and educators should review the information for their chosen features and the [Subprocessors and Service Providers](https://ilmiya.com/sub-processors-list/) register before providing student information to an AI feature. A claim about one provider or feature should not be assumed to cover every model or integration.

Faith-related learning, support needs, or private messages can reveal religious beliefs or health information. Where special-category rules apply, the controller needs an additional legal condition before using those details in an AI feature. A general account approval does not supply it. Ilmiya and the school or educator must assess the risks and complete any required privacy impact assessment, including the effects on children, before the affected processing. Any required consultation about remaining high risk must also be completed.

Keeping information for a chat history, an education record or an agreed archive does not authorize using it to train a model. Only include information needed for the learning task. Follow your school or educator’s rules before uploading student records or another person’s work. The [Global Privacy Policy](https://ilmiya.com/privacy-policy/) explains how requests about access, correction and deletion are handled, including when your school or educator manages the account.

<a id="spoken-interactions"></a>

## Spoken interactions

Voice features process sound so a person can speak to the service and receive a response. Audio can reveal names or other personal details, including speech from people nearby. Transcription can make mistakes, especially with names, accents or background noise.

Allowing microphone access lets your device capture sound. It does not replace any notice, parental authorization or permission from other participants that the institution or Ilmiya must obtain. Start a voice feature only where you have permission to use it, and end the session when you finish. Before a voice activity is enabled for learners, Ilmiya must explain its handling of audio and transcripts to the school or educator, including any recording and retention. The school or educator must give the notices and obtain any additional permission required for the activity. The [School and Institution Consent notice](https://ilmiya.com/parental-consent/) explains the authorization process.

<a id="transparency-and-safeguards"></a>

## Transparency and safeguards

Where the law requires disclosure, Ilmiya must clearly tell a person that they are interacting with AI by their first interaction. The explanation must be accessible and suitable for the learner’s age. Schools and educators must provide the notices their use requires and explain how learners can raise concerns.

The available features and controls depend on the service and configuration provided. Applicable legal safeguards must be in place for the enabled use. Ilmiya and the Customer must assess requirements for privacy by design, including the greater protection children need at different ages.

Our design priorities include limiting unnecessary data, testing for harmful or misleading responses, and making it possible to act on reports. These are ongoing goals. A safety measure cannot ensure that every response is accurate, appropriate or free from bias.

<a id="creating-and-sharing-ai-generated-content"></a>

## Creating and Sharing AI-Generated Content

You must have the rights or other lawful authority needed for material you submit to an AI feature. Review generated material for accuracy and suitability before sharing it with learners or colleagues. Describe AI’s role where disclosure is required by law, the applicable agreement or the educational context, and do not misrepresent authorship. Rights in generated output and pre-existing material follow the MSA and Terms of Service. The standard curriculum license continues to govern supplied learning materials and permitted adaptations.

<a id="ilmiya-in-the-classroom"></a>

## Ilmiya in the Classroom

Educators should explain the capabilities and limitations of AI in language suited to their learners, show how to check an answer, and help students recognize when they need human guidance. Follow the institution’s rules for permitted AI use and avoid submitting unnecessary personal information. Consider whether a response treats people fairly and includes assumptions that are unsuitable for the learner or curriculum.

Staff using an AI feature should remain familiar with its current instructions and the changes relevant to their teaching. Concerns about bias, incorrect information or a missing safeguard can be raised through the reporting route below.

<a id="report-a-problem"></a>

## Report a problem

Tell your teacher or educator if something in your learning activity concerns you. To report an Ilmiya AI issue, email [support@ilmiya.com](mailto:support@ilmiya.com) with the feature name and a brief description. Include a conversation reference if available. Avoid sending a full student record or another person’s sensitive information in the first email; we can arrange an appropriate way to investigate.

The [Support Policy](https://ilmiya.com/support-policy/) explains the support available under your school or educator’s agreement. This statement does not promise continuous monitoring of conversations or an emergency response.

<a id="updates"></a>

<a id="changes-and-contact"></a>

## Changes and Contact

We may update this statement as the service develops. The review date identifies the version you are reading. Changes to binding obligations follow the applicable agreement and law. A revised statement does not remove existing privacy rights or change how previously collected information may lawfully be used.

Where the governing agreement requires advance notice of a material change, including an applicable 30-day notice period, Ilmiya will provide that notice. Questions about this statement may be sent to [support@ilmiya.com](mailto:support@ilmiya.com).

[All legal documents](https://ilmiya.com/legal/)[Contact legal](https://ilmiya.com/legal-inquiry/)

---

[View this page on Ilmiya](https://ilmiya.com/responsible-ai-statement/) · [All pages in Markdown](https://ilmiya.com/markdown-index.md)