Azrivo / Legal / Acceptable Use Policy

Acceptable Use Policy

Effective date: July 20, 2026 · Version 1.4

This Acceptable Use Policy ("AUP") sets the rules for using Azrivo. It forms part of our Terms of Service. We may update it as the service and the risks around AI evolve. If you break these rules, we may remove content, and suspend or terminate your access.

1. Universal prohibitions

You may not use Azrivo to:

  1. Break the law — engage in or facilitate illegal activity, including trafficking of people, drugs, or stolen goods, or infringing others' intellectual-property or contractual rights.
  2. Compromise systems — create or distribute malware, gain unauthorised access to systems or data, or build denial-of-service or intrusion tooling.
  3. Attack critical infrastructure — target power, water, financial, medical, transport, or electoral infrastructure.
  4. Develop weapons — design, build, or deploy weapons, explosives, or biological, chemical, nuclear, or radiological agents.
  5. Promote violence or hate — incite, threaten, or glorify violence, terrorism, or extremism, or harass or demean people on the basis of protected characteristics.
  6. Violate privacy — collect, expose, or process personal data without a lawful basis, conduct surveillance, or impersonate a person or organisation.
  7. Endanger children — produce, solicit, or facilitate any form of child sexual abuse material or the exploitation of minors. This is reported and acted on without exception.
  8. Cause psychological harm — promote suicide or self-harm, eating disorders, or harassment of individuals.
  9. Deceive at scale — generate spam, fraud, phishing, scams, fake reviews, counterfeit documents, plagiarism, or coordinated disinformation.
  10. Abuse the platform — attempt to jailbreak, reverse-engineer, scrape, extract, or replicate the underlying models or system prompts; circumvent rate limits, access controls, or safety measures; or resell access without authorisation.
  11. Misuse a person's likeness or voice — generate realistic, stylised, or animated human-like faces or depictions of identifiable people; perform face swaps; create deepfakes; synthesise or impersonate anyone's voice; or use any person's likeness, image, or voice without that person's explicit consent. Azrivo's image generation exists to produce charts, diagrams, and illustrations for documents — it is not a face, avatar, portrait, or voice tool, and this is enforced: requests for human faces, portraits, avatars, or likenesses are declined at generation time; newly generated imagery is checked for human-like faces before it is saved — the full image for generations, the edited region for region edits; and the region editor will not edit a selected area that contains a face.
  12. Make automated decisions about people — use Outputs to decide, score, rank, or categorise individuals in ways that produce legal or similarly significant effects for them (for example employment, credit, insurance, housing, education, or policing decisions). Azrivo is a thinking aid; it is not an adjudicator, and its Outputs are not a lawful basis for a decision about a person.

We do not permit the generation of sexual content involving minors under any circumstances, and we restrict other sexually explicit content.

2. High-risk uses — no professional reliance

Some Azrivo specialists discuss topics such as legal, medical, financial, or other regulated matters. Their Outputs are not professional advice and are not a substitute for a qualified human professional. You must not:

  • present Azrivo Outputs as professional, licensed, or expert advice; or
  • use Azrivo to make high-stakes decisions (legal, medical, financial, safety, or employment) without independent review by a qualified person.

Every conversation in the Azrivo app carries a standing notice that Azrivo is AI and that its Outputs are not professional advice, and a connected messaging app shows you the same notice when you link it; respect it. Outputs delivered through a connected messaging app are the same AI Outputs, with the same limits.

3. Conduct in debate, panel, plan, and tool use

Azrivo can run multiple AI specialists in parallel, set them debating, build plans, search the web, and generate images, documents, and other artifacts. When using these features you must not:

  • use generated artifacts or images to deceive, defame, or infringe rights;
  • direct tools (such as web search or fetch) at targets you are not authorised to access, or to overload third-party services;
  • automate Azrivo in ways that disrupt the service or other users.

4. Reporting and enforcement

If you see misuse, or content that violates this AUP, contact abuse@azrivo.com. We may investigate, remove content, and suspend or terminate accounts. Serious violations (such as child-safety threats) are reported to the appropriate authorities.

Changelog

  • 1.4 (effective 2026-07-20) — §2 widened: connecting a messaging app now shows the same AI / not-professional-advice notice at link time, and the sentence says so — shipped in the same change as the notice itself.
  • 1.3 (effective 2026-07-20) — §1.11's image-generation sentence now states an enforced control: face/portrait/avatar/likeness requests are declined at generation time, newly generated imagery is face-checked before it is saved (the full image for generations, the edited region for region edits), and the region editor refuses to edit a selected area containing a face — shipped in the same change as the enforcement itself.
  • 1.2 (effective 2026-07-20) — §2's closing line corrected: it claimed a per-specialist regulated-topics disclaimer that the product does not show. It now describes the real control — the standing AI / not-professional-advice notice under every conversation.
  • 1.1 (effective 2026-07-17) — §1 gains two prohibitions: 11. Misuse a person's likeness or voice (human-like faces and depictions of identifiable people, face swaps, deepfakes, voice impersonation, and any use of a person's likeness without consent — Azrivo's image generation is for charts, diagrams, and illustrations, not faces or voices) and 12. Make automated decisions about people (no scoring, ranking, or categorising individuals where it has legal or similarly significant effects).
  • 1.0 (effective 2026-07-07) — Initial published version.