Acceptable Use Policy

1. Scope

This policy applies to organizations using Mthandizi Pro and, where relevant, to the end-users who interact with an organization’s assistant. It supplements the Mthandizi Terms of Service.

2. Content you may not upload or generate

• Content you do not have the right to use — including third-party copyrighted material (e.g. someone else’s book, report, or paywalled article) that you are not licensed to upload.
• Unlawful, defamatory, hateful, or harassing content.
• Personal data you have no lawful basis to process, or special-category data without appropriate safeguards.
• Malware, or content designed to deceive (e.g. impersonation, disinformation).

3. How you may not use the service

• Attempting to access, probe, or retrieve another organization’s data, or to circumvent tenant isolation or security controls.
• Sharing, reselling, or sublicensing API keys, or exceeding/circumventing quota limits.
• Embedding a public key on domains you do not control, or forging request origins.
• Overloading or disrupting the service (e.g. abusive automated traffic).
• Misusing the AI assistant — prompt-injection or jailbreak attempts, trying to extract another tenant’s data or the underlying prompts/models, or using it to generate prohibited content.
• Sending spam or unsolicited bulk messages, including via the SMS or WhatsApp channels, or harvesting contact details for such purposes.
• Using the assistant to provide regulated advice (medical, legal, financial) as if it were a qualified professional, without appropriate human oversight and disclaimers.

4. Your end-users

If you deploy the public widget, you are responsible for your end-users’ conduct through it and for any notices or consents their use may require.

5. Enforcement

We may investigate suspected violations and, where warranted, rate-limit, suspend, or revoke keys or Pro access. Serious or repeated violations may lead to termination under the Terms of Service.