Skip to content
THE CARIBBEAN & LATIN AMERICA ADRIAN DUNKLEY ↗

LAC responsible AI charter

A voluntary working charter for organisations and creators using AI in Latin America and the Caribbean.

Choose one useful task, name the person responsible and agree what the system must never do. Test those boundaries before giving AI access to customers, money or operational systems.

This charter is Adrian Dunkley's proposed practice guide. It is voluntary and does not establish legal compliance, certification or government endorsement. Check the rules that apply in your own jurisdiction.

Make these commitments operational

  1. Define the purpose. Write down the task, who benefits and who could be harmed. Compare the AI option with a simpler manual process before choosing a tool.
  2. Limit data and access. Use only information needed for the task. Check the supplier's storage and training terms before uploading personal, confidential or unpublished material. Give connected tools the smallest permissions required.
  3. Keep accountable human decisions. Identify who can reject an output, stop the service and correct a mistake. For decisions affecting rights, health, employment or essential services, seek the appropriate professional and legal review before deployment.
  4. Test in local conditions. Check the languages, accents and cases your users bring. Include weak connectivity, unavailable data and service interruption. Record failures as well as successes.
  5. Respect people and creative work. Obtain permission for uses of a person's voice or likeness. Check rights to music, images and training or reference material. Do not pass synthetic work off as someone's authentic statement or performance.
  6. Explain and correct. Tell affected people when AI materially shapes a service or communication. Give them an accessible way to raise a concern and reach someone who can change the outcome.
  7. Measure the whole cost. Compare completed, accepted work with the previous process. Include checking, corrections, subscriptions, staff training and service failures. Do not present a pilot's time saving as national economic growth.
  8. Stop and reassess. Suspend a use when it exposes protected information, exceeds its permissions or produces a harmful decision. Keep a manual alternative and review the system after changes to its model, data or task.

Use the pledge in your team

Read the pledge together, then record the task, responsible person, review date and stop conditions in your own project notes. This page does not collect signatures or enrol anyone in a programme.

English

I will use AI for a defined purpose, protect personal information, respect creators’ rights and check outputs before relying on them. I will explain significant AI involvement, keep a person accountable for consequential decisions and provide a way to question or correct a result. I will stop a use that causes harm or exceeds its agreed limits.

Español

Me comprometo a utilizar la IA para un propósito definido, proteger la información personal, respetar los derechos de los creadores y comprobar los resultados antes de confiar en ellos. Explicaré los usos relevantes de la IA, mantendré a una persona responsable de las decisiones con consecuencias importantes y ofreceré una vía para cuestionar o corregir un resultado. Suspenderé cualquier uso que cause daño o exceda los límites acordados.

Português

Comprometo-me a utilizar a IA para uma finalidade definida, proteger as informações pessoais, respeitar os direitos dos criadores e verificar os resultados antes de confiar neles. Explicarei os usos relevantes da IA, manterei uma pessoa responsável pelas decisões com consequências importantes e disponibilizarei uma forma de questionar ou corrigir um resultado. Suspenderei qualquer uso que cause danos ou ultrapasse os limites acordados.

Français

Je m’engage à utiliser l’IA dans un but défini, à protéger les informations personnelles, à respecter les droits des créateurs et à vérifier les résultats avant de m’y fier. J’expliquerai les usages importants de l’IA, maintiendrai une personne responsable des décisions lourdes de conséquences et offrirai un moyen de contester ou de corriger un résultat. Je suspendrai tout usage qui cause un préjudice ou dépasse les limites convenues.

From a pledge to evidence

Before the first pilot, save a short task description, the permissions granted, the test cases and the approval decision. During the pilot, record errors, complaints and correction time. Continue only when the responsible person can explain the results and the remaining risks.

The Safety & alignment toolkit can structure that review. Its scores are screening aids, not independent security tests or regulatory approvals. Creators can use the role-specific guides for rights checks and practical workflows.

Sources and regional policy

The charter translates broad governance principles into proposed working practices. It draws on UNESCO's emphasis on human rights, privacy, oversight and accountability. The detailed task controls and pledge are the author's synthesis.

KEEP LEARNING

PLAYBOOK UPDATES.

Regional guidance, practical skills and project news.

By subscribing you request AI Playbook emails. Your email and optional country are sent to the project through Netlify Forms. For questions or to request removal, contact the project.

Cite this page

Dunkley, Adrian. “LAC responsible AI charter.” LAC AI Playbook. https://aiplaybooklac.netlify.app/responsible-ai/. Include your access date when citing this changing resource.

Geographic coverage: Latin America and the Caribbean. This describes the subject of the guidance, not the author's location or a service address.

Credit Adrian Dunkley when quoting or adapting the playbook. Cite the original research for its findings. Third-party text, photographs and other materials retain their own rights; citation alone does not grant permission to reproduce them.