Watch the recording.

 

“Governance is not a blocker, it’s an enabler.”  That message — voiced by Mathilde Forslund, CEO of Transform Health — surfaced repeatedly during a recent Bay Area Global Health Alliance discussion on artificial intelligence (AI) in global health and reflected a growing recognition that the greatest barriers to scaling AI may have less to do with the technology itself than with the systems underneath it.

As governments, donors, and health organizations race to deploy AI tools, speakers from HealthAI, Transform Health, and Vital Strategies argued that the field risks repeating familiar mistakes from digital health: fragmented pilots, weak public-sector oversight, siloed data systems, and investments that prioritize short-term innovation over long-term system capacity.

On May 7, the Alliance convened leaders from across the AI and global health ecosystem for a candid conversation on what AI governance means in practice — and why it is quickly becoming central to whether AI strengthens health systems or deepens existing inequities. Co-moderated by Daniel Messer, chief digital and IT transformation officer at FHI 360, and Sara Anderson, executive director of the Alliance, the discussion focused not only on safeguards and regulation, but on the deeper infrastructure required for responsible and sustainable AI adoption.

“From my experience at FHI 360, working with various tech partners, AI governance is about making sure we use these tools intentionally — protecting people, data, and trust, while still enabling innovation,” Messer shared, grounding the conversation in a central theme: governance should be an enabler, not a barrier.

But speakers also warned that without stronger governance and oversight, adoption itself could stall. Peiling Yap, chief scientist at HealthAI, described how ministries of health are already being overwhelmed by waves of AI pilots and vendor proposals, often without the regulatory clarity or evaluation frameworks needed to determine which tools can be trusted. “The ultimate question that is holding [ministries] back is who can they trust?” Yap said.

Throughout the conversation, speakers emphasized that some of the most important AI investments may not look like AI investments at all. Pedro do Carmo Baumgratz de Paula, senior vice president of global innovation at Vital Strategies and Alliance board member, pointed instead to interoperable data architecture, civil registration and vital statistics, digitized primary care systems, workforce capability, and sustainable financing as the foundations that ultimately determine whether AI tools can succeed.

The panelists also warned that the funding landscape has yet to catch up to that reality. “They’re still investing in pilots and solutions,” Forslund said, urging donors to invest more seriously in the system strengthening needed for sustainable transformation.

Key Takeaways

• Governance is not a blocker — it is an enabler. Responsible governance creates the trust, safeguards, oversight, and regulatory clarity needed for AI to scale safely and effectively.
• Data governance is central to country ownership. Countries and communities must be able to control, govern, and benefit from their own health data.
• The most important AI investments may not look like AI investments. AI readiness depends on the systems that make responsible use possible — from interoperable data architecture and civil registration and vital statistics to primary care digitization, workforce capability, and long-term financing.
• The funding landscape needs to move beyond “pilotitis.” Donors and partners must invest not only in pilots and solutions, but in the system-strengthening required to make AI tools sustainable, scalable, and embedded in real health systems.
• Responsible AI must ask who benefits. Tools should be evaluated by who they serve, whose data they use, and who may be left invisible.

 

The Safeguards Needed for Trusted AI Adoption

“We strongly believe that when the right safeguards, oversight, and regulatory clarity are not in place, adoption stalls.” – Peiling Yap, HealthAI

Using the analogy of building a house, Yap described AI governance as the architecture needed to make AI safe and usable in health systems. She shared how countries need the architectural blueprints of national AI strategies, the foundation of ethical principles and frameworks, the walls of horizontal legislation, the windows of health sector regulation, and the roof of international standards and guidance.

While AI is advancing faster than many governance systems can respond, Yap emphasized that healthcare remains a regulated domain grounded in the principle of “do no harm.” 

“We strongly believe that when the right safeguards, oversight, and regulatory clarity are not in place, adoption stalls,” Yap said. 

Yap shared how HealthAI is seeing that tension already playing out in Zambia, where the Ministry of Health (MOH) has been overwhelmed by vendors proposing AI pilots. While AI holds real potential to transform care, countries need governance and market authorization processes in place before tools can be safely adopted. “The ultimate question that is holding [ministries] back is who can they trust?” Yap said.

She also pointed to two practical gaps: adaptive AI systems are producing outputs that are “very dynamic and therefore also very difficult to be reproduced,” while existing evaluation methods were built for static products. At the same time, AI governance now cuts across ministries and departments, making cross-institutional coordination critical. To respond, some countries are turning to innovative regulatory approaches such as regulatory sandboxes, model cards, change-control plans, and post-market surveillance to monitor AI across its lifecycle.

 

Moving from Principles to Implementation

“AI is only as trustworthy as the data and governance systems to underpin it, and those systems must be anchored in national decision making.” – Mathilde Forslund, Transform Health

Forslund emphasized that AI governance is not just a technical challenge, but a political one. While commitments and principles on digital health and AI are growing, she said the field is still falling short in turning them into enforceable country-level frameworks.

“Without the right political commitment, without the right systems in place, we won’t be able to achieve the transformation that we all would like to see,” Forslund said.

That implementation gap is especially clear in data governance. As countries adopt digital health tools and AI systems, Forslund warned that guardrails around data ownership, privacy, accountability, and equity remain uneven or missing entirely, highlighting the need for common standards. She pointed to Transform Health’s technical support for the African continental framework on health data governance as one example of efforts to connect regional standards with country-level implementation.

“AI is only as trustworthy as the data and governance systems to underpin it, and those systems must be anchored in national decision making,” she said, emphasizing that national governments and citizens must be able to control and benefit from their own health data if AI is to benefit the public.

With the World Health Assembly approaching, Forslund said the priority is moving “from rhetoric into action” through concrete decisions on frameworks, legislation, resourcing, and technical support that help countries build accountable national systems.

 

AI Readiness Is System Readiness

“AI cannot compensate for weak public health systems. It can only build on what is already institutionally, digitally, and politically in place,” said de Paula pointing to how countries need more than AI tools. 

He shared how countries need digitized and interoperable data systems, public ownership and governance of those systems, workforce capacity, sustainable financing, and legal and regulatory structures that allow AI to translate into public health action.

“AI cannot compensate for weak public health systems. It can only build on what is already institutionally, digitally, and politically in place.” – Pedro do Carmo Baumgratz de Paula, Vital Strategies

De Paula also pointed to work in Brazil where, building on existing electronic medical records and health data systems, Vital Strategies helped create AI-enabled applications to augment the capacity of primary health care units by alerting professionals to potential risks, including gender-based violence and suicide ideation.

However, he emphasized that the visible tool is only possible when the foundations are in place. “The shiny application, the demo, the dashboard that usually involves the decision maker, it relies on a solid foundation,” he said.

He also warned that fragmented systems can leave governments unable to use their own public health data. In one city, a contract with an IT provider prevented the public sector from freely accessing electronic medical record data without extra payment — a reminder that weak foundations can lead to vendor lock-in and loss of sovereignty.

“When you have this type of fragmentation or weakness in capacity, you are unable to advance to the future of what AI can provide to public health,” de Paula said.

 

Moving Beyond “Pilotitis”

Across the discussion, speakers warned that AI risks repeating a familiar pattern in digital health: promising pilots that fail to scale.

Yap emphasized how governance should shape decisions much earlier, including which tools are funded, piloted, and selected for scale. “[Innovators and researchers] start with a pilot but also oftentimes end up in a condition of what I like to call pilotitis,” she noted, where promising tools are tested in small pilots but never move into integrated, sustainable, systemwide use.

Avoiding that cycle requires shifting investment away from “shiny tools” alone and toward the systems that make adoption possible. De Paula noted that some of the most important AI investments may not look like AI investments at all, pointing instead to interoperable data architecture, civil registration and vital statistics, primary care digitization, workforce capability, and long-term financing.

Forslund added that the funding landscape also needs to catch up. “They’re still investing in pilots and solutions,” she said, encouraging donors to really look into the system’s strengthening that is so fundamental for sustainable transformation.

Procurement also emerged as a practical lever for moving from pilots to scale. “You can make procurement a governance instrument,” de Paula said, noting that contracts can require interoperability, auditability, exit rights, local data residency, and other safeguards that help countries retain control over how AI systems evolve.

 

A Call to Build the Systems AI Needs

“Before you invest in the model, ask whether the system underneath is the one that you would still want in five years from now or ten years from now. If it’s not, let’s work on the system.” – Pedro do Carmo Baumgratz de Paula, Vital Strategies

The conversation made clear that responsible AI in global health will not be achieved through better tools alone. It will require the governance, infrastructure, financing, and accountability needed to make those tools useful, trusted, and sustainable.

The call to action is to invest not only in AI models, but in the systems underneath them: stronger data governance, clearer regulatory pathways, interoperable infrastructure, workforce capacity, meaningful coordination, and procurement practices that protect country ownership. Speakers also emphasized that scaling AI responsibly requires asking not only whether a tool works, but who it serves, whose data it learns from, and who may be left invisible.

As Forslund put it, “Governance is not a blocker, it’s an enabler.” Yap echoed that point, urging the field to see governance “as the highways that you need to build for the cars to drive on.”

De Paula offered perhaps the clearest test for the field: “Before you invest in the model, ask whether the system underneath is the one that you would still want in five years from now or ten years from now. If it’s not, let’s work on the system.”

 

Explore related resources and reports from HealthAI, Transform Health, and Vital Strategies to learn more about the governance, data, and system readiness foundations discussed during the convening. Find more on the Alliance’s AI and global health work here.

 

Speaker Bios 

 

Pedro do Carmo Baumgratz de Paula, Senior Vice President, Global Innovation, Vital Strategies; Alliance board member

Pedro do Carmo Baumgratz de Paula is the Senior Vice President, Global Innovation. De Paula has focused his work and research over the past 15 years on policy innovation, urban studies, regulation, and public health. He was in charge of establishing Vital Strategies’ Brazil office, selecting its team and developing strategies for locally relevant projects and practices. Before joining Vital Strategies, de Paula developed a broad experience in academia, consultancy on policy issues, and legal practice. He lectured in law at his alma mater, the Federal University of Juiz de Fora, practiced administrative and regulatory law, and consulted on a variety of regulatory and policy issues. Pedro’s achievements include helping develop a comprehensive road safety plan that will guide the São Paulo’s actions for the next decade, and fostering the improvement of policymaking by the systematic use of data in multiple governmental entities in Brazil. He played a leading role in helping CONASS (national consortium of state departments of health) and other public health authorities in Brazil in developing innovative and rapid data use tools to respond to the COVID-19 pandemic. More recently, Pedro and the Vital Strategies team have assisted the development of pioneering solutions to use data, AI and communication to drive action towards an array of public health priorities such as mental health, gender-based violence, NCDs, among others. He holds an LL.B. (J.D. equivalent) from the Federal University of Juiz de Fora, a M.Sc. and Ph.D in Economic Law and Political Economy from the University of São Paulo’s Law School. Pedro also took a visiting period for his PhD research, focused on state capacity for government innovation in public health, at the Blavatnik School of Government, University of Oxford, during the fall of 2021.

 

Mathilde Forslund, CEO, Transform Health

Mathilde Forslund is the founding CEO of Transform Health – a multi-sector coalition that sets out to harness digital technology and the use of data to accelerate progress towards universal health coverage (UHC). The coalition brings together governments, private sector and civil society to influence and create an enabling environment for digital transformation of health with a focus on equity, inclusion and sustainability. TH has grown from a core network of 7 members to a network of more than 215+ organisations nationally, regionally and globally, with the set up of national coalitions in 6 priority countries (Ecuador, Indonesia, India, Kenya, Mexico and Senegal) and close collaboration with regional networks and multilateral bodies in Asia, Africa & Latin America. Mathilde is a seasoned advocate with experience bringing partners and networks together to deliver on global agendas. Her career spans over 20 years living in Asia, Africa, Europe, USA and currently Switzerland. She leads the TH secretariat and is a board member of HealthAI (the global agency for responsible AI for Health).  She also represents TH as a member of the UHC2030 Task Force and the UNITE Parliamentary Network on Global Health. She is a passionate champion of women’s digital/data rights, inclusion and representation.

 

Dr. Peiling Yap, Chief Scientist, HealthAI

Dr. Peiling Yap is a pharmacist and infectious-diseases epidemiologist by training, with 15 years of experience in the global and public health sectors. As HealthAI’s Chief Scientist, she is responsible for synthesizing knowledge and developing training curricula for regulators and policymakers to strengthen their capacity to regulate AI in health, leveraging globally defined Responsible AI standards, processes, and tools. She is also leading the establishment of HealthAI’s Community of Practice, which aims to facilitate knowledge sharing and collaboration among diverse stakeholders on implementing regulatory mechanisms and global standards for Responsible AI in health.  She was recently appointed as co-editor-in-chief of Artificial Intelligence in Global Health, a new journal from the Royal Society of Tropical Medicine and Hygiene launching in May 2026. This journal will be dedicated exclusively to advancing research and discourse on AI in global health, with a particular focus on ensuring relevance to low- and middle-income countries. She is also a Board member of the African Population Health and Research Center (APHRC) based in Nairobi, Kenya. Dr. Yap has spearheaded impactful public health research and initiatives globally. At the Swiss Tropical and Public Health Institute, she investigated psychosocial health and nutrition in children and the effectiveness of health education targeting children in China and South Africa. At the National Center for Infectious Diseases in Singapore, she implemented and evaluated interventions for infectious diseases and trained healthcare professionals.

 

Daniel Messer (co-moderator), Chief Digital and IT Transformation Officer (CDxO), FHI 360

Daniel Messer is the Chief Digital and IT Transformation Officer (CDxO) at FHI 360, where he leads the organization’s AI strategy, digital transformation, and enterprise technology modernization. His work focuses on responsible and practical applications of artificial intelligence, including governance and AI‑enabled decision‑making across global health, development, and humanitarian programs. Daniel partners closely with senior leadership to ensure AI investments are ethical, scalable, and aligned with organizational mission and impact, while integrating emerging technologies such as digital financial systems and blockchain into enterprise operations.

 

Sara Anderson (co-moderator), Executive Director, Bay Area Global Health Alliance

Sara Anderson is the first Executive Director of the Bay Area Global Health Alliance, serving in the role since 2020. With more than 30 years of experience in global health, international development advocacy, nonprofit management, partnership development, thought leadership, and strategic communications, she has a proven track record of driving impact. Before leading the Bay Area Global Health Alliance, Anderson was at the forefront of advocacy efforts in pioneering movements, including global surgery, raising awareness about debilitating burn injuries in low-income countries, and addressing the famine in North Korea. Her work on Capitol Hill and with various campaigns and nonprofits focused on shaping U.S. government policy and building strategic partnerships to influence policy and change lives. She holds a master’s degree from Georgetown University and has furthered her education through Stanford’s Designing for Social Systems program and the Maxwell School’s Transnational NGO Leadership Institute at Syracuse University. Additionally, she has served as a coach and partner in Stanford’s d.school Design for Extreme Affordability program.