Abstract

Artificial intelligence (AI) is transforming healthcare, but its rapid deployment raises concerns about equity, transparency, and accountability. Without proper oversight, these systems can reinforce biases, disproportionately affecting marginalised communities. Current regulations and policies fail to fully address these risks, making proactive safeguards essential to prevent systemic health inequities. To address these challenges, we organized the Health AI Systems Thinking for Community (HASTC) workshop at the University of Toronto (October 2024). This cross-disciplinary workshop convened 66 participants from post-secondary, healthcare, and nonprofit sectors to collaboratively discuss the management of AI harm in healthcare. Participants, guided by mentors, analyzed real-world cases of algorithmic bias, privacy risks, and unintended consequences of AI-assisted decision-making. Semi-structured discussions within groups focused on accountability, transparency, and fairness. Through structured discussions, participants identified worst-case scenarios and proposed safeguards at different levels, with implications toward government regulations, institutional policies, and healthcare practices. Key findings emphasized theneed for adaptive, context-specific regulations and discussions to ensure responsible AI use in healthcare. There is a need for ongoing dialogue and reflection. By integrating community-driven advocacy and interactive learning, HASTC highlights the importance of creating AI systems that are fair, accountable, and transparent, to benefit all patients, not just a privileged few.