A trans woman from Bahawalpur – South Punjab sits alone in her room, her eyes fixed on her phone as she waits for an update from a donor. She is hoping for funding for a project that would help provide food to 50 transgender people, a modest effort to support a community already struggling with poverty and exclusion. As the minutes pass, she turns to another concern. She begins searching online for information about nearby health services where she can access hormone replacement therapy. Then her phone beeps.
She reaches for it, expecting an update from the donor. Instead, she finds an abusive message. Then another. And another. Within seconds, her phone is buzzing relentlessly. Her inbox fills with rape threats, death threats and other messages of abuse. A social media post she had shared to raise awareness about food support for transgender people has been reported. Another post was taken down after it was flagged for violating community standards. Yet, hateful comments directed at her remain online.
She became one of the many targets of the wave of digital abuse that came amid the controversy surrounding Pakistan’s Transgender Persons (Protection of Rights) Act, 2018, in 2023, when transgender rights became the subject of intense public and political debate. For transgender people like her, the backlash did not remain confined to television screens, political statements or social media debates. It spilled directly into their inboxes, phones and lives.
This is, however, not an isolated experience. The constant and baseless abuse and violence against them discourages many trans people from sharing their stories, reporting violence, or engaging in public discussions. Some opt not to be online altogether, and others avoid talking about topics directly relevant to their rights and safety. If fear determines who has the floor and who doesn’t, digital platforms stop being places for equal participation.
Online presence matters
For trans people in Pakistan, the internet is not only a space for communication or leisure. It has become an important support system in a society where many are still left out from traditional economic, social and institutional systems. When family rejection, intolerance and violence limit access to opportunities, digital spaces often become one of the few places where transgender people can seek information, build connections and find acceptance, especially transgender individuals living in southern Punjab and other unreached areas in Pakistan.
Alongside, family rejection remains one of the earliest forms of isolation experienced by many transgender people like I did till the age of 21 when finally my family accepted me. Many are forced to leave their homes, lose economic support or live without access to education and healthcare. In such conditions, online communities can provide something that is often missing offline that is a sense of belonging. A simple online contact can lead someone to a support group, a legal source, a healthcare provider or another person who recognizes their lived experience.
Through my development work as founder and CEO of PINK Center in South Punjab, I have experienced how digital spaces can help transgender people access opportunities and support that is not available in their neighbourhood, families or close physical spaces. A lot of transgender community members have used digital networks to find right information about legal aid, emergency resources, educational opportunities, healthcare services among others.
Digital spaces have also revolutionised the way grassroots activism works. Community organisations record human rights abuses such as the recent incident of a young trans girl in Multan who was harassed by a group of boys, and she took them live on social media, arranged emergency responses– like an older transwoman generated Rs35,000 for her kidney treatment in Bahawalpur through fund raising appeals on her social media statuses– facilitate awareness raising and link survivors of violence to legal and social services. In times of crisis such as floods that are very common in south Punjab and other humanitarian situations, the Internet has provided fast information to volunteers, donors and partner organisations where physical access has been limited.
The same spaces which promise greater inclusion are increasingly intensifying exclusion. Automated moderation systems, recommendation AI models and mechanised content review tools frequently struggle to understand the circumstances of transgender people, particularly in countries like Pakistan where gender identities, languages and cultural contexts differ noticeably from the data on which many of these systems are built. Instead of protecting vulnerable communities, digital technologies can often silence their voices, reduce their visibility and leave them more exposed to online abuse.
Difficulties in online verification
Unlike internationally accepted public figures who may build their validity through large-scale public acknowledgment, many grassroots trans activists depend on standard identity certification systems to build trust with journalists, donors, human rights organisations and the communities they serve. But for transgender activists in Pakistan, this process becomes another point of exclusion.
Platforms often require government-issued identification to verify an account, yet many transgender people use a preferred name that differs from the name on their legal documents. Updating official identity records can be a lengthy, expensive and inaccessible process, leaving many unable to meet platform requirements. As a result, activists working on urgent issues such as documenting violence, coordinating emergency responses and advocating for justice may struggle to verify their accounts, reducing their visibility and credibility in digital spaces. A certification system designed to enhance trust can unknowingly consolidate the structural hurdles that transgender communities already face offline.
AI discrimination
AI, on the other hand, is increasingly influencing people’s attitudes towards the digital world. Automated technologies touch every aspect of what people can share, who they can reach and how loud their voices will be, from content moderation to account authentication to search results. These systems are, however, NOT neutral. They bear information, assumptions and theoretical frameworks on which they are based. The experiences of some communities can be misunderstood, or overlooked when they aren’t included in technology development and training data.
This is a massive hurdle for transgender people, especially in South Asian countries such as Pakistan. Numerous AIs designed to moderate content struggle to comprehend gender identities, cultural imagery, regional languages and Hijra Farsi (an Indigenous language of transgender people across South Asia). Such systems tend to be built in a context in which gender is predominantly conceptualised within narrow categories and where much of the digital data is derived from a dominant language and context. Consequently, they may not be aware of the realities of communities which do not correspond to these frameworks.
Pakistan has a long history of a diverse plurality of genders and communities like Hijra, Khawaja Sira and other local gender identities with their cultural significance and social histories. Many digital systems employ simple classifications that cannot always be seen as these identities. Local words, images and communications can be lost in translation for automated tools with no context. For instance, discussions about transgender identity, bodies or human rights are interpreted by automated systems as violations, which potentially leads to the removal or confinement of content. The visibility of community organisations raising awareness about transgender issues can be lessened if the content is wrongly flagged or reported. In some instances, false reports may also result in restrictions or suspension of accounts, with a negative impact on the ability of activists and organisations to reach the communities they serve.
It’s not like AI is targeting racial groups in its quest for equality. But, the deeper problem is that technology can learn from what already exists and when it learns from existing patterns, the experience of marginalised groups is not part of these patterns then, what it learns can reinforce the patterns of marginalisation. A system that isn’t able to distinguish between hate speech against transgender people and education about transgender rights can actually be a system that protects the wrong voices, and silences the voices that need to be protected.
When AI is creating the public face, the trusted faces and the voices people hear online, it also needs to acknowledge communities outside of the core of Pakistan’s biggest cities as Pakistan does not mean Lahore, Karachi and Islamabad, the majority of transgender population don’t live in these three big cities. If not, the most vulnerable users of the Internet may be getting the least out of it.
Developing inclusive AI is NOT just about fixing algorithms. It involves multiple communities in technology design, expands the data used by technology to incorporate various languages and cultural experiences, and extends the involvement of people impacted by technological systems in designing the technology. If this exclusion is not made, AI can just be another obstacle for communities already struggling to gain visibility and recognition.
Amplification of Hate Speech
Online abuse does not only refer to individual acts of victimisation when it comes to transgender in Pakistan. How many digital platforms suggest and share content can exacerbate the issue. The aim of a recommendation system is to optimise user interaction with the content based on its appeal and engagement. These systems are not made to spread hate, but content that is emotionally charged and controversial brings more engagement and consequently spreads damaging narratives.
This can be a digital harm cycle for transgender communities. An article in support of trans rights can become the target of abusive posts and spam within a matter of hours. Supportive articles about trans rights can be targeted by abuse or organised abuse and reporting within hours. The more people engage with the post, the more the discussion becomes visible, and a larger audience is exposed to the person/organisation. In the absence of informed discussion, the platform can serve as a space to reinforce prejudice and harassment and to invite more attention to it.
Some trans activists in Pakistan say they spend as much as writing educational material on trans issues as they do in moderating abusive comments and blocking accounts. Community members have also described another form of online abuse. Some transgender activists receive unwanted private messages with sexual advances. When these advances are rejected the harassment can escalate into public attacks. Individuals may begin leaving abusive comments on advocacy posts, questioning the activist’s character, spreading misinformation or portraying their human rights work as “socially unacceptable”.
Reporting these incidents does not necessarily result in positive consequences. When others are posting content that is harmful for the community, but education posts, awareness campaigns or discussions about transgender issues find themselves limited in number or visibility, or are hard to find, community members are often frustrated. These can be achieved through automated systems, through user reporting or through human review or a mixture of all three; regardless of how the outcomes are arrived at, the users who are targeted by abuse may feel less protected than the users who are responsible for the abuse.
There’s been a lot of investment by technology companies in online safety, but safeguarding the welfare of marginal communities is not just about deleting individual posts. It requires systems to be more aware of the nature of coordinated harassment, more responsive to abuse targeting vulnerable populations and less likely to impact negatively on people raising awareness about human rights through these systems.
