This Prompt, Output and Training Data Consent (“Consent”) explains the choices you will be asked to make concerning the collection and use of information submitted to or generated by HUMAIN M3 through Node. Your choices are recorded only when you select the relevant controls in the Node access interface; viewing this page does not constitute consent. Node means HUMAIN's online platform, interfaces, accounts, any enabled application programming interface and related functionality through which HUMAIN makes HUMAIN M3 available. HUMAIN M3 or Model means the experimental generative artificial intelligence model made available through Node. The Preview is made available by AL-MUSTAQBAL LIL-THAKA AL-IISTINAI COMPANY (HUMAIN), commercial register number 1009089438, at King Fahd District 3254, King Fahd Branch Street 7295, Postal Code 12262, Riyadh, Kingdom of Saudi Arabia (HUMAIN, we, us or our).
READ THIS BEFORE CONSENTING. Every prompt and every response is recorded. Authorized automated systems, HUMAIN reviewers and model technology provider reviewers may examine any interaction. If you separately consent, HUMAIN and its model technology provider may select and use prompts, responses and feedback to evaluate, train, fine-tune and further develop HUMAIN M3 and its related safeguards. Do not submit personal, confidential, privileged, sensitive or regulated information.
This Consent applies to eligible users worldwide where HUMAIN makes the Preview available. It supplements the Node – HUMAIN M3 Limited Preview Privacy Notice, which provides fuller information about legal bases, recipients, international transfers, security, rights and regional requirements. Capitalized terms not defined here have the meanings given in the Terms of Use or Privacy Notice.
1. Information Covered
For this Consent:
Inputs include every prompt, instruction, message, uploaded file or content if enabled, API request, edit, rating, reaction, report and other material submitted through Node as part of the Preview.
Outputs include every response, completion, answer, image, code, classification, refusal, warning and other content generated or returned by HUMAIN M3.
Associated Data include your account or API identifier, model and safeguard configuration, timestamps, request identifiers, token counts, session information, feedback, safety signals, IP address, device or browser information, errors and related technical logs.
Every Input and Output is recorded and associated with the relevant account or API credential. HUMAIN may use automated tools to classify, filter, secure and analyze these records.
2. Recording and Review
HUMAIN records and may review Inputs, Outputs and Associated Data to:
- provide, administer, secure and troubleshoot the Preview;
- detect misuse, fraud, security incidents and violations of the Preview documents;
- evaluate accuracy, quality, safety, Arabic and multilingual performance, regional or cultural alignment and safeguard behavior;
- investigate reports, reproduce errors and respond to support, legal or regulatory requirements; and
- measure performance, conduct benchmarking and determine how HUMAIN M3 should be corrected or improved.
Review is not limited to automatically flagged interactions or selected samples. Any interaction may be reviewed where access is authorized and appropriate for the disclosed purposes.
3. Training and Further Development
With the separate Training and Development Consent in Section 8, every Input and Output is eligible for selection. HUMAIN may use selected Inputs, Outputs, ratings, corrections, reports and feedback to:
- create, curate, label, filter and evaluate datasets;
- train, fine-tune, benchmark, test and further develop HUMAIN M3 and its related variants;
- develop, test and improve safeguards, classifiers, monitoring systems and evaluation methods; and
- improve Arabic and multilingual quality, safety and regional or cultural alignment.
HUMAIN may exclude interactions from training, apply quality or safety filters and remove or transform identifiers. Selected material retained as training data will be effectively de-identified where required by applicable law and where reasonably possible. Information incorporated into trained model parameters may not remain attributable to you and may not be technically capable of being extracted, isolated or reversed.
HUMAIN will not treat this Consent as covering advertising, sale of personal data, cross-context behavioral advertising or a materially different purpose. Marketing choices are separate.
4. Who May Access the Information
Access is limited according to role and need to know. Recipients may include:
- authorized HUMAIN personnel and contractors responsible for product, safety, security, legal, privacy, support, evaluation or machine-learning work;
- HUMAIN's contracted model technology provider and its authorized reviewers, which may receive and use Preview data for inference, review, safety, evaluation, training and further development of HUMAIN M3 and its related safeguards, subject to applicable contractual, confidentiality, access-control and data-protection requirements;
- approved hosting, cloud, security, analytics, communications, identity-verification and support providers; and
- professional advisers, auditors, competent authorities or other recipients where permitted or required by law.
The model technology provider may use Preview information for the disclosed inference, review, safety, evaluation, training and development purposes relating to HUMAIN M3 and its related safeguards, in accordance with HUMAIN's provider arrangements and applicable law.
5. Retention and International Processing
| Information | Retention approach |
|---|---|
| Raw user-linked Inputs and Outputs | Ordinarily retained for twelve (12) months from collection, then securely deleted or effectively de-identified unless longer retention is required by law or for a documented legal claim. |
| Selected effectively de-identified training data | May be retained longer for evaluation, safety, training and further development where re-identification is not reasonably possible. |
| Account, verification, consent and security records | Retained for the Preview and afterward as necessary for administration, consent evidence, security, legal compliance and claims under HUMAIN's approved retention schedule. |
| Trained model parameters | May be retained as part of HUMAIN M3 and related systems. Statistical learning incorporated into parameters may not be technically reversible or attributable to an individual interaction. |
HUMAIN and approved service providers may process information in Saudi Arabia and other countries identified in the Node interface or a layered notice. HUMAIN will apply transfer measures required by applicable law. Do not consent or submit an interaction until the applicable location disclosure has been presented to you.
6. Your Responsibilities
You must not submit personal data or confidential, privileged, sensitive, classified or regulated information. You must not submit another person's information unless you have a lawful basis, authority and any required notice or consent.
For API access, these consents are linked to your verified account and apply to calls made using your API credentials. You must not share an API key or use it to provide HUMAIN M3 to third-party end users. Acceptance by an organization representative does not by itself constitute consent on behalf of another individual whose personal data is submitted.
This Consent does not change ownership of Inputs or Outputs, waive the Preview's use restrictions, or authorize production, customer-facing, multi-tenant or unlawful use.
7. Research Access
Research Access is a separate, voluntary option available through the Playground and, where enabled, the API. If you request it, HUMAIN must verify and approve you, and you must separately accept the Voluntary Research Access Addendum and heightened-risk acknowledgement before activation. Research Access interactions are subject to this Consent and may receive heightened review because HUMAIN's filtering-layer safeguards may be reduced or removed.
Declining or withdrawing only the Research Access opt-in does not by itself end eligible Guarded Access. It does not withdraw the recording, review or training choices recorded below.
8. Required Electronic Consents
Before Node enables access, you will be asked to give Consent A and Consent B separately through two unticked controls in the Node interface. Selecting one will not select the other. Neither consent is given by accepting the Terms of Use, viewing this page or merely using Node.
Consent A - Prompt and Output Recording and Review
I consent to HUMAIN recording every Input and Output I submit or receive through Node as part of the HUMAIN M3 Limited Preview, together with related account and technical data. I understand that authorized HUMAIN personnel, automated systems and authorized service-provider reviewers may review any interaction for service administration, safety, security, quality, alignment, support, compliance and evaluation as described in this Consent and the Privacy Notice.
Consent B - Training and Further Development
I consent to HUMAIN and its model technology provider selecting and using my Inputs, Outputs, ratings, corrections, reports and feedback to create and evaluate datasets and to train, fine-tune, benchmark, test and further develop HUMAIN M3, its related variants and safeguards. I understand that the model technology provider may receive and use this information for those purposes and that trained model parameters may not be technically reversible.
9. Access Confirmation
By selecting Continue after activating both controls, you confirm that you have read this Consent and the Privacy Notice, understand that every interaction is recorded, and understand that both consents must remain active for you to submit new Guarded Access or Research Access interactions.
10. Choice, Refusal and Withdrawal
Participation in the Preview is voluntary. You do not have to give either consent. Because recording, review and model development are stated purposes of this limited preview, HUMAIN will not enable or continue new interactions unless both required consents are active. Refusal does not affect any unrelated HUMAIN service.
You may withdraw either consent at any time through the account or privacy control identified in the Preview. Withdrawal must be as easy as giving consent and will disable new Guarded Access and Research Access interactions. Withdrawal does not affect processing lawfully completed before withdrawal, information that HUMAIN must retain by law or for a documented claim, effectively de-identified data that can no longer reasonably identify you, or statistical learning already incorporated into model parameters where reversal is not technically possible.
Where applicable law does not permit HUMAIN to offer the Preview on this consent basis, HUMAIN may withhold access in that jurisdiction or present an approved regional notice and control before access is enabled.
11. Changes and Contact
HUMAIN will request fresh affirmative consent before using identifiable information for a materially new purpose that is not compatible with the choices recorded here. A change to the Terms of Use does not by itself amend this Consent.
Questions, consent withdrawal and privacy-rights requests may be submitted through the privacy channel identified in the Node interface or Privacy Notice. You may also contact HUMAIN at [email protected].