“The future of AI in HIV should be shaped with communities, not simply built from data about them. That is how we ensure AI becomes a tool for shifting power, not another mechanism for concentrating it.” – Solange Baptiste, ITPC
“The question is not whether AI belongs in HIV programming. It already does.” That was the message delivered by Izukanji Sikazwe, the Global Fund’s Head of HIV, during the first-ever AI-focused plenary at an International AIDS Conference. At AIDS 2026, AI moved closer to the center of the HIV conversation: from what AI could do for HIV to how it is already being tested and used; from what is being learned about its benefits and risks to how stakeholders can differentiate evidence of use from evidence of impact.
More recently, ITPC Executive Director, Community Working Group Co-Lead of the Lancet Commission on AI + HIV, and AVAC Board Member Solange Baptiste pushed the conversation further: “Who controls the data that makes AI possible, and who benefits from it?” Drawing on lessons from the HIV movement’s long fight for community leadership and equity, she highlights risks of data extraction and calls for stronger governance, meaningful consent, human oversight and accountability, “The future of AI in HIV should be shaped with communities, not simply built from data about them.”
Across these conversations, the field is exploring AI-supported HIV prevention and care, demand generation, community engagement and the issues of privacy, harm and governance. “We are well and truly in the AI age, and the question is: How are we going to use it for greater impact?” Dr. Sikazwe asked at the conference.
AVAC’s second quarterly issue of AI and HIV looks across these conversations: from important perspectives like Baptiste’s, to data from AIDS 2026, to the emerging tools, policy developments and research that will shape how AI is developed, governed and used across HIV and global health.
AI at AIDS 2026
Research and conversations at AIDS 2026 explored how AI should be governed, who benefits and how it can be used in real-world HIV prevention and care. Following are highlights from the presentations shared in Rio:
Harnessing Artificial Intelligence for HIV: Breakthrough or Buzzword?
In the first-ever AIDS conference plenary focused on AI, Izukanji Sikazwe from the Global Fund explored how AI is already being used in the HIV response and what it will take to move promising innovations beyond pilots to sustainable, scaled programs. She argues that AI should be built into health systems as shared infrastructure for public good, with community leadership, equity, privacy, human oversight, governance and sustainable financing embedded from the start. She closed with a reminder that success should be measured by infections prevented, and people kept in care, not tasks automated.
AI health coaches in South African public clinics
Population Services International (PSI), Wits Health Consortium and Audere Africa presented early data on an AI health coach designed to provide health information and support, with a particular focus on HIV and TB. Developed through a participatory, human-centered design process and introduced across 161 public clinics in South Africa, early data of the Coach Mpilo program showed the value of hybrid models that combine AI’s reach and accessibility with human oversight. In the study, clinicians, health promoters and community health workers were far more effective at referring new users to the AI health coach than print materials or social media.
Rethinking PrEP scale-up through everyday conversations
Researchers in South Africa explored whether AI could use everyday health conversations to identify people who may benefit from long-acting PrEP. Rather than answering questions about PrEP, the approach considers whether AI could help programs recognize prevention needs and generate demand among people who might otherwise never seek out PrEP information. The study showed that if questions are designed around the decisions they inform, self-care conversations can help target limited supply of prevention tools with those most likely to benefit.
From chat to care: AI conversations as community intelligence
Another study led by ITPC Global explored whether aggregated AI health conversations could provide real-time insight into the questions communities are asking, the barriers they face and the services they need. An important consideration is whether this data can strengthen community-led monitoring and program responsiveness without turning people’s health conversations into another form of surveillance.
Building safety into AI for HIV and sexual-health self-care
Researchers also looked at the risks of misinformation and the need for quality assurance in AI-generated HIV and sexual and reproductive health (SRH) information. One study tested a safety system for AI-powered HIV and SRH self-care tools in South Africa and Zimbabwe using 1.5 million+ AI-mediated interactions. They built a layered system that continuously detects potential harms, responds automatically, alerts clinicians when necessary and incorporates human review and follow-up – demonstrating why always-on monitoring is essential.
Artificial Intelligence in the HIV response: All the good stuff from AIDS2026
An analysis by Jirair Ratevosian of the Duke Global Health Institute looked comprehensively at AI-related abstracts, key themes from use cases, and emerging trends at AIDS 2026. His review shows the potential of AI to support HIV testing and prevention, including PrEP, and improve linkages to care and services, particularly in sub-Saharan Africa. “AI is becoming a new front door to HIV services,” he writes, but the analysis also shows that while most of the research was focused on feasibility and pilot implementation of AI, a significant evidence gap around health outcomes, as well as privacy, community engagement, and governance remains. He concludes that: AI is working, but trust is what makes or breaks the product; specialized tools beat generics; human-in-the-loop is becoming standard; the field needs to move from pilots to proof of outcomes; and scale depends on systems, not just models. See more on research presented at AIDS 2026 below.
Governance, Equity, and Accountability
In this powerful piece, ITPC Executive Director and AVAC Board Member Solange Baptiste asks a fundamental question about the growing use of AI in HIV: “Who controls the data that makes AI possible, and who benefits from it?” Drawing on lessons from the HIV movement’s long fight for community leadership and equity, she examines the risks of data extraction and calls for stronger governance, meaningful consent, human oversight and accountability. As she writes, “The future of AI in HIV should be shaped with communities, not simply built from data about them.”
A new report from WHO reinforces the need to introduce AI safely, equitably and effectively within existing health systems. These implementation choices will help determine whether AI strengthens care or creates new risks and inequities. The report examines what it takes to move AI from promising pilots into routine practice, including governance, infrastructure, workforce readiness and real-world implementation.
AI Technology
Anthropic launched Claude Science, an AI application (app) designed to help scientific and pharmaceutical research and development (R&D) by integrating the tools and packages that researchers most use. The tool is intended to help researchers navigate the complexity of biology and accelerate scientific discovery to develop specialized AI tools for biomedical research rather than general-purpose models.
South African researchers evaluated Your Path, an AI-powered conversational tool designed to link people to HIV prevention or treatment services after self-testing. In a study of 100 community members and 25 health care providers, the tool was viewed as easy to use, private and nonjudgmental, and it increased confidence in accessing HIV services, though it did not significantly change those with already high intentions to seek care. Some participants raised data privacy concerns, while healthcare providers noted limitations in the tool’s emotional responsiveness and ability to accommodate language differences.
Research
The UN Independent International Scientific Panel on AI released its first preliminary report, offering a global, independent assessment of AI’s opportunities, risks and impacts. Developed by 40 experts from across the world, the report aims to provide a shared evidence base for governments navigating rapidly evolving AI technologies and will inform the Panel’s first comprehensive report in 2027.
The second edition of the Global Index on Responsible AI (GIRAI) assessed how 135 countries and jurisdictions are governing AI across five areas, including inclusion, ethics, labor, trust and safety, and public-sector use. Going beyond whether responsible AI policies exist, the 2026 Index examines whether those policies are being implemented and delivering results.
Policy Developments
WHO published a new discussion paper that examines how AI is changing the way health evidence is synthesized and used in policymaking. The paper examines the intersection of AI and EIP, outlining how AI can support problem identification, policy design and implementation through enhanced data integration, predictive modelling, scenario simulation and adaptive feedback. It also cautions about bias, transparency, accountability, overreliance on AI-generated evidence, and unequal access.
What We’re Reading
- As AI fears mount, Gates hopes to make it a force for good in global south—Devex
- The turbulent AI era is here. The choices we make now are critical—Gates Notes
- Global Index on Responsible AI — 2026 Report, Global Index on Responsible AI
- UN Agencies Release Publication on AI-Enabled Health Innovation and Intellectual Property—World Intellectual Property Organization
- From Breakthrough to Follow-Through—A Public Health Agenda for AI—JAMA
- BFI #2: One Billion Doctors—With & For Substack
- Investing in AI-enabled machines to improve early TB detection in western Kenya—PATH
- Generating synthetic multi-national longitudinal cohorts for clinically grounded HIV research—Nature Communications
- How AI responds to common HIV/AIDS questions: ChatGPT versus DeepSeek— Frontiers in Public Health
- As AI fears mount, Gates hopes to make it a force for good in global south—Devex
- Africa Can Grow Faster With AI—If It Moves Now—International Monetary Fund
- AI company Anthropic announces it will begin developing drugs of its own—STAT
- Researchers caught in the crossfire as companies and government grapple over AI safety—Science
- Kenya plan to sell citizens’ data draws criticism—Semafor
- Kenyan scientist wins Sh187m grant to predict disease outbreaks—Daily Nation
- From Market Shaping to Market Building: What Happens When the Old Global Health Playbook No Longer Applies?—Bay Area Global Health Alliance
- AI’s Biggest Risk Isn’t the Technology — It’s Who Controls It, Raffi Krikorian Tells Alliance Annual Meeting—Bay Area Global Health Alliance
- AI promises judgment-free sexual health advice. What could go wrong?—Devex
- Balancing Algorithmic Capabilities and Human Empathy to Build a Sustainable Smart Health Service Model for HIV—Mediamedic
- Use This Simple AI Framework to Stop Drowning in AI Governance—ICTWorks
- U.S. Health Officials Move Quickly to Deploy Medical A.I. Despite Concerns—The New York Times
- There Is Something We Have to Do Right Now About A.I.—The New York Times