AbteeX AI Labs
Draft policy · 26 April 2026

Responsible use.

AbteeX AI Labs builds research systems, model infrastructure, and agentic operations software from Aotearoa New Zealand. Responsible use means capability is paired with consent, review, provenance, privacy, and clear limits on deployment.

1. Human authority stays visible

AI systems should support people who remain accountable for decisions. AbteeX systems should not be positioned as autonomous replacements for clinicians, lawyers, public officials, safety operators, financial advisers, or other professionals where human judgement and legal responsibility are required.

  • High-impact recommendations should be reviewable before action.
  • Users should be able to inspect the evidence and context behind important outputs.
  • Systems should make uncertainty, limitations, and failure modes visible.

2. Data must have rights, context, and restraint

Data is not neutral simply because it is available. Personal information, clinical material, commercially sensitive records, Māori knowledge, community-held knowledge, and local datasets require explicit rights and appropriate safeguards before use.

  • Do not submit data unless you have authority to use it.
  • Do not use AbteeX systems to infer, expose, or reconstruct sensitive personal information without a lawful and ethical basis.
  • Do not treat mātauranga Māori, community knowledge, or culturally significant material as generic training content.

3. Prohibited use

AbteeX systems and materials must not be used to cause harm, bypass safeguards, or conceal accountability.

  • No malware, credential theft, phishing, exploit automation, or unauthorized access.
  • No deception, impersonation, fraud, spam, or manipulation of vulnerable people.
  • No unlawful surveillance, profiling, or discrimination.
  • No medical, legal, financial, employment, housing, education, or public-sector decisions without appropriate human review and compliance controls.
  • No generation or distribution of abusive, exploitative, or non-consensual content.

4. Agentic systems need operational controls

Agents that can call tools, update records, send messages, or affect business processes need stronger controls than chat interfaces. AbteeX expects agentic deployments to record intent, evidence, proposed action, approval state, result, and rollback path.

Principle: if a system can act, the organization should be able to explain why it acted, who authorized it, what changed, and how to reverse or correct the result.

5. Reporting concerns

Security, privacy, safety, or misuse concerns can be sent to [email protected]. Include enough detail for review, but avoid sending unnecessary personal or sensitive information.