MedSpir AI

1: Being able to come together as 'a global forum' where all the parties come to agree upon 'core principles' for 'AI' such as 'to be honest, be accountable, to be fair, and to have human intervention. 2: To be able to establish mechanisms by which to hear the minority voices as well at a minimum to hear from the innovators and practitioners from developing nations so that the governance of 'AI' reflects the many different realities and challenges that exist in place. 3: To be able to provide each country, through workforce and knowledge building opportunities, both 'AI' literacy, regulatory knowledge, and the need to develop technical infrastructure for deployment. 4: Establishing best practice standards for interoperability and responsible AI deployment that will promote standardisation for safety, privacy and ethics across international borders and in critical sectors (such as healthcare). 5: Willingness to monitor continuously and improve, thereby evolving the governance of 'AI' to keep it in sync with new technologies and the impact of AI on society, as such, this outcome will provide the global voice of all partners, establish a global forum for cooperation on 'AI' risk mitigation and maximize benefits of AI (for global innovation/investment).

• Safe, secure and trustworthy AI
• AI capacity-building
• Interoperability of governance approaches
• Transparency, accountability, and human oversight

There are four main areas of focus where MedSpir AI needs to make the biggest impact in the regions we serve. 1: The Importance of Safe, Secure, and Trustworthy AI: We must ensure that AI tools are safe, reliable and unbiased (to protect human lives) and this is crucial in health care. 2: Building Up Capacity for AI: We need to develop local expertise and technical skills so that AI can be adopted ethically and sustainably in a way that is locally relevant. 3: Interoperable Governance Approaches: By creating harmonised standards and encouraging cross-border collaboration, we will avoid fragmentation and help ensure that AI solutions can develop smoothly, quickly and safely. 4: Transparent and Accountable Oversight of Human Activity: Explainable models of AI are essential to maintaining user, policy and community trust through mechanisms of accountability. Together, these four areas will recognise both the technical and ethical issues related to the deployment of AI in addition to providing practical guidance to governments, developers and civil society on how they can manage AI in an effective and actionable manner. Furthermore, they establish a basis for an inclusive and actionable governance system that finds an acceptable balance between innovation and being a societal protector.

1: Equity and inclusion in AI benefit: Governance frameworks need to specifically address how AI advancement will not further contribute to existing inequality and inequity across all areas. Special emphasis should be given to the impact of AI on healthcare, education, and digital access. 2: Sector-specific ethical guidelines: Every sector that has a direct, substantial impact on people (health, finance, education, and others) will require separate governance as a means of addressing the unique risk and consequences of each sector to society. 3: Data sovereignty and privacy: Governance must provide both countries and individuals with continued ownership over their data to ensure protection from misuse. In addition, there must be a means for countries/individuals to have the ability to securely share their data for the development of AI. 4: Sustainable development of AI: The energy consumption of many AI models raises environmental concerns due to the production of excessive energy resources. Frameworks need to provide incentives for more efficient and responsible design and implementation of AI. These issues address the requirement for holistic, future-oriented, and globally-relevant governance for AI and will promote trust, inclusion, and sustainable development outcomes.

Challenges: 1: Limited regulatory frameworks: Many countries lack clear policies for AI safety, accountability, and ethical use, especially in healthcare. This creates uncertainty for innovators like MedSpir AI and slows adoption of life-saving AI tools. 2: Capacity constraints: There is a shortage of skilled personnel to develop, deploy, and regulate AI responsibly, limiting the ability to scale healthcare innovations effectively. 3: Data infrastructure and interoperability: Fragmented health data systems and inconsistent standards make it difficult to deploy AI tools across hospitals, research institutions, and government programs. 4: Transparency and trust issues: Without clear governance, clinicians and patients may hesitate to use AI, slowing innovation and affecting healthcare outcomes. Opportunities: 1: Inclusive policy development: Engaging local innovators and healthcare practitioners ensures policies are practical, context-specific, and accelerate adoption. 2: Capacity-building initiatives: Training clinicians, developers, and policymakers strengthens the workforce needed to deploy trustworthy AI solutions. 3: Regional collaboration and standards: Harmonizing AI guidelines across African countries improves interoperability, safety, and trust, attracting investment and supporting sustainable innovation. 4: Ethical, patient-centered AI: Africa can lead in deploying AI that prioritizes equity, safety, and explainability, setting global best-practice examples.

1: Setting shared principles and standards: The Dialogue can establish globally recognized norms for trustworthy, ethical, and safe AI deployment, creating a common reference for policymakers and practitioners. 2: Facilitating knowledge exchange: By sharing best practices, lessons learned, and regulatory approaches, countries and organizations can accelerate learning and avoid repeating mistakes. 3: Bridging geographic and sectoral divides: The Dialogue can amplify voices from low- and middle-income countries and underrepresented sectors, ensuring AI governance frameworks are globally relevant and inclusive. 4: Promoting coordinated action: The Dialogue can catalyze multi-stakeholder initiatives that address cross-border AI challenges, such as data privacy, interoperability, bias mitigation, and ethical deployment in critical sectors like healthcare.

The AI Dialogue should build upon and connect with initiatives such as the OECD AI Principles, UNESCO's Recommendation on AI Ethics, the Global Partnership on AI (GPAI), and regional AI strategies across Africa, Asia, and Europe. These platforms provide valuable guidance on ethics, safety, and technical standards. Gaps remain, however, in bringing these efforts together in a forum that is truly inclusive of private sector innovators, civil society, and practitioners from developing regions. Many existing mechanisms focus primarily on policy, leaving a gap in actionable insights from real-world deployment and sector-specific experience. The AI Dialogue can add value by: 1: Integrating diverse perspectives: Ensuring that regional and sector-specific insights, especially from health and other high-impact industries, are included in global governance discussions. 2: Facilitating actionable recommendations: Moving beyond principles to practical guidance, testing, and pilots that inform both policy and implementation. 3: Encouraging collaboration and partnerships: Linking governments, technical communities, and the private sector to jointly address challenges like interoperability, safety, and ethical deployment. 4: Monitoring emerging developments: Providing a platform to assess new AI technologies, risks, and cross-cutting issues in real time, allowing governance frameworks to evolve proactively. By connecting existing efforts with underrepresented voices and sectoral expertise, the AI Dialogue can become the go-to hub for practical, inclusive, and forward-looking international cooperation on AI governance.

1: Governments can share regulatory experiences and challenges in integrating AI into national health systems, helping to shape practical governance models that work across contexts. 2: Private sector innovators, especially health‑tech startups like MedSpir AI, provide on‑the‑ground lessons from deploying clinical AI tools, including safety issues, integration challenges, and real‑user feedback. 4: Clinicians and healthcare practitioners can speak directly to risks and benefits encountered in patient care, ensuring governance reflects what happens at point‑of‑care. 5: Academia and research institutions contribute evidence on bias, model performance, and clinical outcomes. 6: Civil society and patient advocacy groups ensure the patient voice and public trust are included. Format & structure recommendations: 1: Sector‑specific breakout groups: e.g., Health AI, Education AI, Public Services AI, with outputs tailored to real challenges and use cases. 2: Case‑study sessions: where healthcare AI implementers (like MedSpir AI) present problems, solutions, and learnings. 3: Hybrid consultations and pre‑event surveys: to include voices from clinics and communities that can't travel. 4: Interactive panel and working sessions: to convert high‑level principles into actionable healthcare governance recommendations. 5: Post‑Dialogue working groups: to follow through on commitments, especially for health sector standards.

1: Frontline clinicians and health practitioners who work with AI tools daily and can highlight risks that don't show up in labs. 2: Healthcare technology startups from Africa and other low‑ and middle‑income regions, often excluded from global policy conversations despite solving real problems. 3: Patients and caregivers, whose lived experience with health AI tools must inform trust, usability, and risk frameworks. 4: Health data stewards and hospital administrators, who grapple with privacy, interoperability, and safety in real systems. How to include them: 1: Dedicated healthcare streams within the Dialogue that require representation from clinical implementers, patients, and regional health‑tech innovators. 2: Virtual engagement platforms with targeted outreach to healthcare workers and patient groups. 3: Pre‑Dialogue consultations and focus groups specifically for health AI stakeholders, feeding findings into the main event. 4: Travel support or sponsorships for participants from underrepresented regions or organizations.

1: Healthcare AI Case Clinics: Short, structured sessions where startups and hospitals present real deployment challenges (e.g., bias in diagnostic tools), and multi‑stakeholder groups co‑design governance solutions. 2: Interactive Simulation Labs: Participants explore real governance trade‑offs using hypothetical healthcare AI incidents, surfacing blind spots and consensus areas. 3: Hybrid Participation Platform: Real‑time polls, Q&A, and shared drafting tools allow clinicians, patients, and remote innovators to influence outputs live. 4: Post‑Dialogue Working Pods: Small teams (government, private sector, clinicians) tasked with building implementation roadmaps for health AI governance aligned with Dialogue outcomes. 5: Health AI Innovation Showcase: A space to highlight responsible, ethical, explainable AI tools, with feedback loops from regulators, practitioners, and patients.

1: Regulatory frameworks with sector-specific guidance: Countries like the UK (NHS AI Lab) and Canada (Health Canada guidance on AI in medical devices) provide clear, risk-based standards for safe and ethical AI deployment in healthcare. These frameworks guide innovators like MedSpir AI to design tools that meet safety, privacy, and accountability requirements. 2: Ethical guidelines and principles: The OECD AI Principles and UNESCO Recommendation on the Ethics of AI emphasize transparency, fairness, and human oversight. In healthcare, these principles ensure AI tools support clinicians and patients without introducing bias or compromising safety. 3: Data governance platforms: Platforms such as OHDSI (Observational Health Data Sciences and Informatics) and FHIR standards for health data interoperability allow secure, standardized sharing of health data for AI model development while respecting privacy. These approaches enable MedSpir AI to deploy AI solutions across hospitals and clinics with consistent data quality and safety. 4: Open collaboration and sandbox models: Regulatory sandboxes, such as the UK's NHS AI sandbox or Singapore's AI governance sandbox, allow innovators to test AI tools in controlled environments, iteratively improving models and governance practices before wide-scale deployment. 5: Capacity-building initiatives: Training programs for clinicians, policymakers, and developers, like WHO's Digital Health and AI Capacity-Building Toolkit, promote responsible AI use, ensuring governance aligns with practical needs in healthcare settings.