Pro Gender Support Foundation Pakistan

Policies

Artificial Intelligence (AI) Policy

How Pro-Gender Support Foundation Pakistan uses artificial intelligence responsibly, safely and in the service of the communities we work with.

PGSF uses artificial intelligence to extend what a small, committed team can do for transgender (khawaja sira) persons and other marginalised communities across Pakistan. Used well, AI helps us write, translate, analyse, plan and communicate faster, so that more of our time goes to the people we serve.

AI also brings real risks. It can produce false information, reflect bias against the very communities we work with, expose personal data and be used to deceive. This page sets out the highlights of our AI Policy: what we allow, what we never do, and who is accountable. We will update it as technology and the law develop.

People stay accountableA named person reviews and approves every AI-assisted output.
Community data is protectedInformation that could identify a community member never goes into an AI tool.
Nothing fake, everWe never invent stories, quotes, photographs or results.

What is artificial intelligence?

Artificial intelligence refers to software that can generate, analyse, translate, transcribe, summarise or predict content. It includes AI assistants and chatbots, image, audio and video generators, translation and transcription services, and AI features built into everyday tools such as email, word processing and social media. Our policy covers all of these, whether free or paid.

The policy applies to our Board of Directors, staff, Transgender Community Steering Committee, Technical Advisory Group, volunteers, interns, consultants and partners acting on behalf of PGSF.

Our commitment to responsible AI

PGSF permits and encourages the responsible use of AI across all of its teams. We see AI as an assistant that supports human judgement, not a replacement for it.

Human oversight

Every piece of work produced with the help of AI remains the responsibility of the person who uses it and of PGSF. AI never makes final decisions about people.

Transparency

We are open about our use of AI wherever it matters to the people reading, viewing or relying on our work, and we never present AI-generated content as something it is not.

Do no harm

We assess the risk to community members before using AI and choose the safer option when in doubt. Safeguarding always comes before efficiency.

Our guiding principles

Our approach draws on the UNESCO Recommendation on the Ethics of Artificial Intelligence, the OECD AI Principles, the NIST AI Risk Management Framework, ISO/IEC 42001, UNICEF’s Policy Guidance on AI for Children and the Core Humanitarian Standard, adapted to the context of Pakistan.

  • Human oversight and accountability for every AI-assisted output.
  • Privacy and data protection for community members, staff, volunteers, donors and partners.
  • Fairness, inclusion and dignity, checking outputs for stereotypes and bias.
  • Accuracy and quality, verifying facts, figures and sources.
  • Community voice: AI never replaces genuine consultation with the community.
  • Security and lawfulness, using organisational accounts and complying with the laws of Pakistan.
  • Responsible resource use, choosing free and secure tools and using AI where it adds real value.

How our teams use AI

Every PGSF team may use AI as a co-worker that supports its own work, always under human direction and review. Typical uses include:

  • Programmes: co-drafting training materials and concept notes, co-summarising research and co-translating general information into Urdu and other languages.
  • Communications and advocacy: co-drafting and co-editing posts, articles and website content, and co-producing captions and subtitles.
  • Fundraising and partnerships: co-researching opportunities, and co-drafting proposals, reports and presentations, following each donor’s own rules on AI.
  • Monitoring, evaluation and research: co-analysing anonymised data and co-drafting reports from verified findings.
  • Website and digital: co-writing and checking website copy, and co-improving accessibility and search visibility.
  • Legal affairs and human rights: co-summarising public laws and judgments, always checked by a qualified person.
  • Administration, finance and governance: co-drafting letters, minutes, procedures and policies for human review.

Uses that need approval

Some uses carry higher risk and need written approval from our President and Chief Executive Officer before they begin:

  • Any AI tool that interacts directly with community members or the public, such as a chatbot. People will always be told they are talking to an automated system and will always be able to reach a person.
  • Analysing programme data that relates to individuals, even after anonymisation.
  • Adopting a new AI tool or connecting one to PGSF accounts.
  • Producing AI-generated audio or video for publication.

What we never do

  • Enter names, CNIC numbers, contact details, photographs, gender identity, health or HIV status, case histories or locations of community members into AI tools.
  • Let AI decide who receives support, who is selected for a stall, or who is hired.
  • Create deepfakes, or realistic images, video or audio of real people without their informed consent.
  • Invent or alter testimonials, case studies, quotes, photographs, statistics or results.
  • Use AI to impersonate anyone or to mislead donors, partners, authorities or the public.
  • Use AI for surveillance, facial recognition or to profile anyone’s gender identity, religion, health or other sensitive characteristic.
  • Rely on AI for legal, medical or psychosocial advice without review by a qualified professional.
  • Use AI where a donor or partner has prohibited it.

Protecting personal data

Before using any information with an AI tool, our team classifies it as public, internal, confidential or highly sensitive. Highly sensitive information, meaning anything that could identify a community member, survivor or child, or any health, identity, legal case or safeguarding information, is never entered into an AI tool. We use organisational accounts, keep an internal register of approved tools and switch off data training wherever possible. Our Privacy Policy and Safeguarding Policy apply in full to all AI use.

Images, stories and content labelling

  • The stories of the people we work with belong to them. We tell real stories only with informed consent and never use AI to invent them.
  • AI-generated images, illustrations, audio and video that we publish are clearly labelled as AI-generated.
  • We never create AI images that depict community members, staff or partners as if they were real photographs.
  • Photographs of real people receive only basic edits, such as cropping, colour correction or blurring faces for protection.
  • Where AI has played a substantial part in a publication, we say so.

Safeguarding and AI-facilitated harm

Transgender persons and other marginalised groups are frequent targets of online abuse, including fake images, voice cloning and manipulated content. PGSF treats AI-facilitated harm as a safeguarding concern. We support affected people, help preserve evidence and, with their consent, assist them in reporting the matter to the platform concerned and to the relevant authorities under the Prevention of Electronic Crimes Act, 2016. No information about children is ever entered into AI tools.

Accountability and review

Our Board of Directors approves the AI Policy and oversees its implementation. Our President and Chief Executive Officer is accountable for it across PGSF and appoints an AI Focal Person who maintains the register of approved tools, provides training and receives reports of concerns. Everyone covered by the policy receives an introduction to it during induction and refresher learning every year. Any mistake or misuse must be reported within 24 hours so that harm can be limited, and people who raise concerns in good faith are protected. We review the policy at least once a year.

Full policy

To request a full copy of PGSF’s AI Policy, please email info@progendersupport.org.

This policy will continue to be updated as technology and the law in Pakistan develop. We welcome your feedback at info@progendersupport.org. Last updated: November 2026.

Scroll to Top