As artificial intelligence becomes a practical reality in regulatory work, a fireside chat at the AI in Regulation Conference examined AI not simply as a set of tools, but through the lens of institutional readiness. The session, From compliance to capacity, examined how regulators across sectors are beginning to build the internal foundations required for safe and effective AI adoption.
Moderated by Jennifer Quaglietta, CEO and Registrar of Professional Engineers Ontario, the discussion focused on why traditional compliance frameworks alone are no longer sufficient for governing fast-moving, complex technologies. Quaglietta framed the challenge as one of regulatory adaptation, noting that many oversight systems were designed for predictable environments, while AI operates within dynamic ecosystems shaped by continuous interaction between technology, people, and institutions. The key question, she suggested, is not whether AI is coming, but how regulators engage with it without undermining public trust.
Drawing on experience at the Municipal Property Assessment Corporation (MPAC), Soussanna Karas reflected on how her organization is embedding AI into strategic planning and starting with low-risk, internal use cases. Rather than deploying public-facing systems, MPAC focused on internal tools and staff-led pilot ideas to build familiarity, confidence, and governance discipline.
Karas emphasized that capacity-building is as much cultural as it is technical. Successful AI adoption, she argued, depends on leadership alignment, workforce education, and clear accountability, not just software solutions. MPAC’s internal pitch process, which paired bottom-up experimentation with structured governance review, helped surface realistic use cases while managing risk.
Data readiness emerged as a recurring theme. Karas cautioned that AI initiatives often fail when data quality, documentation, and ownership are unclear, reinforcing that early governance decisions frequently determine long-term outcomes. Transparency and human accountability, particularly where AI supports regulatory judgment rather than replaces it, were also highlighted as essential.
Rather than offering a blueprint, the session surfaced enduring questions for regulators: How ready are boards and staff? Which decisions should never be automated? And how can AI strengthen, rather than erode, professional competence and public trust?
