Between Statistical Invisibility and Surveillance: Data, Gender Identity, and the Right to Health
VIEWPOINT
Sebastian León-Giraldo, Juli Salamanca, and Lina Quevedo
Introduction
The fight over gender is also a fight over data. Anti-gender politics operate through laws, speeches, and restrictions on care.[1] But they also reach census categories, health records, internet searches, social media feeds, applications, and the datasets used to train artificial intelligence. These systems do more than describe reality. They shape which health needs become visible, what information people find, how institutions understand their bodies, and who has the power to define them. At a time when democracy, science, and public health are under attack, defending rights also means defending the informational conditions that make those rights visible and enforceable.
This is also a right to health issue. Under international human rights law, health services must be available, accessible, acceptable, and of good quality.[2] For transgender and gender-diverse people, these dimensions increasingly depend on data infrastructures. Services are harder to plan when needs are absent from official statistics. Acceptability is undermined when records erase names and identities. Quality is compromised when clinical tools rely on incomplete or biased information, including data shaped by binary or pathologizing assumptions. And accessibility is weakened when digital environments return misinformation, hostility, or silence. Across these different infrastructures, we identify three interconnected threats: statistical invisibility and erasure, distortion, and extraction and surveillance.
Statistical invisibility and erasure
These processes are related, but they do not work in the same way. Statistical invisibility occurs when systems have never collected or correctly classified the information needed to recognize particular populations and health inequities. Erasure occurs when an existing capacity for recognition is removed or restricted. In the United States, approximately 360 federal data collections removed at least one measure of sexual orientation or gender identity between January 2025 and January 2026; 83% of these removals were processed as “non-substantive” administrative changes.[3] A variable can disappear without a clinic closing or a right being formally repealed, but its disappearance can still affect policy, budgets, research, and accountability.
In Colombia, the problem takes a different form. Our research on death certificates showed how the longstanding omission of transgender identities undermines accurate mortality data, dignity, mourning, collective memory, and the ability to recognize mortality and violence as public problems.[4]
We call the response to both invisibility and erasure statistical justice: the right to be recognized in government systems, and for official data to reflect the true diversity of societies.[5] This does not mean that every system should collect gender identity, or that collecting more data is always better. The questions are more basic: Is the information needed for a legitimate health or public purpose? Are the categories accurate? Have communities had a role in their design? Does collecting the information lead to resources or accountability? And is it protected against uses that may cause surveillance or harm?
Distortion
Transgender people often look for health information online because providers may lack knowledge, services may be difficult to access, or clinical encounters may feel unsafe. Online spaces can connect people with competent providers and community knowledge, but they can also expose them to censorship, misinformation, and hate speech.[6] For health, then, the question is not only whether information exists. It is also what search engines rank, what recommendation systems amplify, and what generative artificial intelligence produces when someone asks about hormones, fertility, mental health, bodily changes, or gender identity.
A recent study testing four large language models with medical prompts found inappropriate responses in all four, with prompts mentioning LGBTQIA+ identities showing more severe forms of bias.[7] The World Health Organization has also warned that generative artificial intelligence in health can produce false, inaccurate, biased, or incomplete information and may be trained on poor-quality or biased data.[8] When available data and online texts reproduce pathologization or political disinformation, automated systems can reproduce these narratives at scale and present them with the authority of a seemingly neutral answer.
At the same time, digital spaces can provide recognition, community, and support. This matters because responses to online harm cannot simply tell transgender and nonbinary people to disengage from these spaces.[9]
Extraction and surveillance
Transgender people can remain statistically invisible, or be actively erased from systems of public recognition, while being overexposed elsewhere. Governments, health institutions, platforms, and applications can collect information that reveals intimate aspects of health and identity without adequate safeguards, clear limits on secondary uses, or meaningful control by the people concerned.
A participatory study across Colombia, Ghana, Kenya, and Vietnam found that about three-quarters of focus group participants described technology-facilitated abuse. Participants also raised concerns about third-party data sharing and government surveillance. In Colombia, transgender sex workers described harms that moved from online spaces into their offline lives.[10] Major platforms also collect and monetize extensive amounts of personal data.[11] But these risks are not limited to commercial actors. Information collected by public institutions for health, identification, or social protection can also become harmful when unintended actors gain access to it, when databases are linked without adequate safeguards, or when information is reused for surveillance, discrimination, or exclusion.
Data sovereignty helps describe what is missing here: meaningful power for people and communities over how data about them are collected, classified, accessed, interpreted, shared, and reused. Health data governance principles similarly emphasize protection against discrimination and surveillance, informed consent, community participation, and equitable representation.[12] For trans communities, being recognized cannot come at the cost of being exposed to uses of information that they cannot anticipate, challenge, or control.
Conclusion
The “fight for rights” must therefore include a fight over data infrastructures. Human rights organizations should pay attention when gender-related measures needed to identify health inequities are absent, removed, or changed, and ask governments and institutions to justify those decisions. Researchers and communities can audit search engines and artificial intelligence systems using trans health questions in multiple languages. Regulators should require transparency, avenues for redress, and safeguards for sensitive data. Health institutions, meanwhile, should involve trans communities as decision-makers in the systems that represent them. A regional observatory on gender data and health could help preserve evidence when institutions retreat, compare patterns across countries, document algorithmic harms, and support community-led responses.
The current backlash can keep trans people statistically invisible, erase them from systems in which they have begun to be recognized, distort information about their health, and expose them through systems they do not control. Statistical justice, algorithmic accountability, and data sovereignty are therefore connected conditions of the right to health. Defending that right means defending health care, as well as the data environment through which needs are recognized, knowledge is produced, decisions are automated, and lives become legible to power.
Sebastian León-Giraldo, PhD, is a postdoctoral researcher at Universidad de los Andes, Bogotá, Colombia, and a researcher at Liga de Salud Trans, Bogotá, Colombia.
Juli Salamanca is executive director and cofounder of Liga de Salud Trans, Bogotá, Colombia.
Lina Quevedo is coordinator of the accompaniment and pedagogical model at Liga de Salud Trans, Bogotá, Colombia.
Please address correspondence to Sebastian Leon-Giraldo. Email: sd.leon10@uniandes.edu.co.
Competing interests: None declared.
Copyright © 2026 León-Giraldo, Salamanca, and Quevedo. This is an open access article distributed under the terms of the Creative Commons Attribution-Noncommercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
References
[1] A. Perez-Brumer, N. Valdez, and A. I. Scheim, “The Anti-Gender Threat: An Ethical, Democratic, and Scientific Imperative for NIH Research/ers,” Social Science and Medicine 351/Suppl 1 (2024).
[2] Committee on Economic, Social and Cultural Rights, General Comment No. 14, UN Doc. E/C.12/2000/4 (2000), para. 12.
[3] L. J. A. Bouton and E. Redfield, Removal of Sexual Orientation and Gender Identity from Federal Data Collections: January 2025 to January 2026 (Williams Institute, UCLA School of Law, 2026).
[4] S. León-Giraldo, P. Vargas, J. Salamanca, et al., “When Death Tears You Apart: The Omission of Transgender Identities in Colombian Death Certificates,” International Journal of Transgender Health 27/4 (2026).
[5] S. León-Giraldo, F. Sánchez Osorio, N. Chalela, and F. Álvarez, “When Populations Are Not Counted: Statistical Justice and Gender-Identity Data in Colombia’s Official Statistics” (manuscript under review at Data and Policy, 2026).
[6] L. Augustaitis, L. A. Merrill, K. E. Gamarel, and O. L. Haimson, “Online Transgender Health Information Seeking: Facilitators, Barriers, and Future Directions,” Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems (2021).
[7] C. T. Chang, N. Srivathsa, C. Bou-Khalil, et al., “Evaluating Anti-LGBTQIA+ Medical Bias in Large Language Models,” PLOS Digital Health 4/9 (2025).
[8] World Health Organization, Ethics and Governance of Artificial Intelligence for Health: Guidance on Large Multi-Modal Models (2025).
[9] S. M. Coyne, E. Weinstein, J. A. Sheppard, et al., “Analysis of Social Media Use, Mental Health, and Gender Identity among US Youths,” JAMA Network Open 6/7 (2023).
[10] Digital Health and Rights Project Consortium, Paying the Costs of Connection: Human Rights of Young Adults in the Digital Age in Colombia, Ghana, Kenya and Vietnam (2025).
[11] Federal Trade Commission (United States), A Look Behind the Screens: Examining the Data Practices of Social Media and Video Streaming Services (2024).
[12] Transform Health, Health Data Governance Principles (2022).