This Acceptable Use Policy (“Policy”) governs all access to and use of HUMAIN M3 through Node as part of the Preview, including the Playground, any enabled API and activated Research Access. It is issued 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.
GLOBAL SCOPE. This Policy applies wherever HUMAIN makes the Preview available. You must comply with the laws applicable to your location, activity, Inputs, intended use and treatment of Outputs. HUMAIN may restrict countries, persons, organizations, networks and use cases.
LIMITED PREVIEW. HUMAIN M3 is experimental and not production-grade. Outputs may be wrong, unsafe, offensive, biased, unlawful, culturally inappropriate or non-unique and safeguards may fail. Every prompt and response is recorded. Separate affirmative electronic consent is required for disclosed review, training and further development.
This Policy forms part of the Terms. The Model Terms, Provider Requirements Register and Privacy Notice also apply. The Research Access Addendum creates a narrow exception only for activated, voluntary Research Access and cannot override applicable law.
You accept this Policy through the same unticked electronic acceptance control used for the Terms, with separate links to both documents and an acknowledgement of the Privacy Notice. Mere access or use does not replace a required affirmative acceptance or consent.
1. Purpose and Preview Status
Use the Preview only for lawful evaluation, testing, feedback and non-production prototyping. Do not use it as a production service, expose it to third-party users, rely on it operationally or deploy it in a regulated, high-impact or safety-critical activity.
The Preview may be changed, rate-limited, suspended or withdrawn at any time. It provides no access to model weights, training datasets, hidden prompts, system configurations or source code.
2. Eligibility and Access Modes
You must be at least eighteen (18), legally capable of accepting the Terms and eligible under applicable law and HUMAIN's access controls. Provide accurate identity, age, country, organization and contact information and cooperate with verification.
Guarded Access. The Playground and any enabled API operate with HUMAIN's filtering-layer safeguards enabled by default. Keep accounts and keys secure; do not share, transfer, resell or expose them or use them to serve unauthorized third parties.
Research Access. Any eligible adult user may voluntarily opt in and request Research Access through the Playground and, where enabled, the API. HUMAIN verifies and selects participants and may apply session, category, concurrency, volume and rate caps. HUMAIN may reduce or remove its filtering-layer safeguards. No academic credential or formal test plan is required. Access begins only after an activation notice.
3. Permitted Evaluation
You may evaluate Arabic and multilingual quality, dialects, code-switching, reasoning, safety, safeguards and regional or cultural alignment; test controlled non-production concepts; and provide ratings, feedback and reports within your enabled access mode.
You are responsible for every Input, your treatment of every Output and all activity under your credentials. An approval, Output or technical capability is not a certification, endorsement or permission to act.
4. Inputs and Restricted Data
Submit only material you are lawfully entitled to provide and for which required notices, permissions and consents exist. Use synthetic or fictitious information wherever possible.
Do not submit personal data or confidential, privileged, sensitive, regulated, classified, export-controlled or security-critical information, including:
- health, biometric, genetic, financial, criminal, children's or other sensitive personal data;
- identity documents, national identifiers, payment data, passwords, authentication information, security answers or live access tokens;
- classified, defence, national-security, government-confidential or critical-infrastructure information;
- trade secrets, confidential business information, privileged communications, unpublished source code or professional-secrecy material;
- another person's data, likeness, voice or communications unless submission is lawful, necessary and expressly enabled by HUMAIN; or
- live-system details, real credentials, malware, exploit code or material creating an immediate security or safety risk.
5. Responsible Use of Outputs
Independently verify every Output for accuracy, completeness, legality, appropriateness and fitness for purpose. Apply qualified human judgment before communicating, publishing, implementing or acting on it.
Do not present an Output as factual certainty, professional advice, an official determination, exclusive content or human-authored content where that would be misleading. Preserve required AI labels, provenance notices, watermarks and safety markings.
Do not follow, execute, share or operationalize harmful instructions. Do not open links, run code, contact targets, access systems, acquire materials or take real-world action based on unsafe Output.
Neither HUMAIN nor an underlying model technology provider is responsible to you for an Output or an action taken in reliance on it, to the maximum extent permitted by law. Mandatory rights and non-excludable liability remain unaffected.
6. Prohibited Uses for Every User
You must not use the Preview, directly or indirectly, to:
- violate or facilitate violation of law, regulation, court order, governmental direction, sanctions, export control, contractual obligation or third-party right;
- endanger national security, public safety or public order, or promote, recruit for, finance or facilitate terrorism, violent extremism, criminal activity or related organizations;
- create, request, possess, distribute or facilitate child sexual abuse material, sexual exploitation, grooming, non-consensual intimate imagery or sexual content involving minors;
- plan, enable or materially facilitate violence, murder, abuse, trafficking, self-harm, suicide, weapons, military targeting or chemical, biological, radiological or nuclear harm;
- generate, deploy or facilitate malware, ransomware, spyware, credential theft, phishing, unauthorized access, unlawful interception, denial of service, intrusion, data theft or system compromise;
- conduct fraud, scams, impersonation, identity theft, deceptive synthetic media, evidence fabrication, spam, unlawful gambling, market manipulation or coordinated inauthentic behavior;
- target, harass, threaten, defame, discriminate against, dox, surveil or exploit a person or group, invade privacy or unlawfully infer sensitive attributes;
- create or disseminate false, manipulated or synthetic content likely to deceive, defraud, harm or unlawfully influence individuals, markets, public institutions or elections;
- infringe intellectual-property, privacy, publicity, confidentiality, database or other rights, or submit material you lack authority to use;
- make a solely or materially automated high-impact decision, provide professional advice, or operate in healthcare, emergency response, critical infrastructure, law enforcement, criminal justice or another regulated or safety-critical setting;
- resell, sublicense, redistribute, publicly deploy or provide the Preview to unauthorized third parties, or use an API in a customer-facing, multi-tenant or production service;
- scrape or systematically extract Outputs; reverse engineer, decompile or disassemble the Preview; extract weights, hidden prompts, system configuration, secrets or other users' data;
- use the Preview or Outputs to train, benchmark or support a competing model, dataset, product or service;
- interfere with, overload or damage the Preview; evade monitoring; defeat limits or controls; conceal a violation; falsify identity, authority, age or location; or
- cause HUMAIN or an underlying provider to breach law or contract, or create an unacceptable legal, safety, security, privacy or operational risk.
7. Applicable Provider Requirements
The current Provider Register identifies binding downstream-relevant restrictions for use, content, safety, security, access controls and model protection. It supplements this Policy and the more restrictive applicable requirement controls within that scope.
The Register does not incorporate provider commercial, data-use, intellectual-property, warranty, liability, governing-law or dispute terms and creates no direct provider-user contract or provider enforcement right. HUMAIN will version material updates and request renewed electronic acceptance where required.
8. Safeguards and Research Exception
Guarded Access users must not bypass, disable or systematically probe for evasion of safeguards, monitoring, access controls or technical protections. Incidental failures should be reported.
An activated Research Access participant may intentionally probe otherwise restricted model behaviour solely for confidential safety, security, quality or alignment evaluation, within the activated access parameters. No formal test plan is required. The exception exists only after voluntary opt-in, verification, acceptance, risk acknowledgement, consent and activation.
The research exception does not permit unlawful possession or distribution; real-world harm; targeting actual persons; testing live or third-party systems without the owner's express authorization; executing code or commands; acquiring harmful materials; contacting targets; operational use; publication; or access by an unapproved observer.
Research prompts, Outputs, screenshots, findings, evaluations and configuration details are confidential and may be disclosed only to HUMAIN through the designated channel unless HUMAIN authorizes otherwise. Reducing or removing HUMAIN's filtering-layer safeguards may expose participants to extreme or psychologically distressing content.
9. Recording, Review, Training and Consent
Every Input and Output is recorded with relevant account and technical data. Authorized HUMAIN and provider reviewers may review any interaction under confidentiality, access-control and data-protection obligations. The model technology provider may receive and use Preview data for inference, review, safety, evaluation, training and further development of HUMAIN M3 and its related safeguards, in accordance with HUMAIN's provider arrangements and applicable law.
With separate affirmative electronic consent, HUMAIN and its model technology provider may use selected Inputs, Outputs and feedback to evaluate, benchmark, train, fine-tune and further develop HUMAIN M3 and its related safeguards, in accordance with HUMAIN's provider arrangements and applicable law.
Raw user-linked interactions are ordinarily retained for twelve (12) months. Selected effectively de-identified data may be retained longer for evaluation, safety, training and development. The Privacy Notice explains recipients, locations, retention, rights and withdrawal.
The electronic controls for contractual acceptance, logging/review consent, training/development consent and the Research Access opt-in or heightened-risk acknowledgement must be distinct and unticked. Withdrawal affects future consent-based processing and ends new interactions where processing is integral.
10. Monitoring, Enforcement and Reporting
HUMAIN may use automated systems and authorized human review to evaluate safety and quality, detect misuse and threats, investigate violations and protect users, the Preview and the public.
HUMAIN may warn, remove or restrict content, throttle or end sessions, preserve records, disable credentials, suspend or terminate access, notify an organization and report matters to authorities or affected persons where permitted or required. It may act immediately.
Promptly report harmful or unlawful Outputs, safeguard failures, vulnerabilities, privacy concerns, unauthorized access and imminent harm through the relevant interface channel. Include only the minimum information necessary.
11. Intellectual Property and Model Protection
HUMAIN and its licensors retain all rights in Node, the Preview, Model, software, algorithms, interfaces, documentation, safeguards, branding and related technology. No licence to weights or underlying technology is granted.
You retain rights you have in Inputs, subject to the Preview documents and consents. Outputs may be non-unique, unprotectable, inaccurate or infringing. You are responsible for assessing and clearing any use.
12. Changes, Suspension, Law and Contact
HUMAIN may update this Policy, the Provider Register or technical controls. Material contractual changes require renewed affirmative electronic acceptance, and processing changes require fresh consent where applicable. HUMAIN may suspend affected access pending implementation.
The governing law, disputes, language, disclaimer, liability and general provisions in the Terms apply. Mandatory local rights remain unaffected.
Questions and reports may be submitted through the channel identified in the Preview interface or Privacy Notice, or directed to [email protected].