Notes below follow the live discussion flow. Timestamps are approximate; quoted segments are verbatim where marked, otherwise close paraphrase from the audio.
--- **[Panel](https://ispe.org/conferences/2025-annual-meeting-expo/agenda#na25ceoct1-gs15):** Tina Kiang, PhD — FDA Tala Fakhouri, PhD, MPH — Parexel International Ian Rees — MHRA (UK) Kevin O’Donnell — HPRA (Ireland) Sarah Pope Miksinski, PhD — Gilead Sciences, Inc. Roger Nosal — Vaxcyte Jeong Yeon Kim, PhD — Ministry of Food and Drug Safety, Republic of Korea Shanshan Liu — No Deviation Pte Ltd --- ### Early Career Moments (≈ 00:01) > “Ian, tell us about a pivotal moment early in your career.” **Ian Rees** reflected on his early participation in ICH during the mid-1990s, describing how it shifted his view of regulation from adversarial oversight to collaboration-first engagement. He shared that the pandemic vaccine response demonstrated what such collaboration could achieve on a global scale. The broader reflection emphasized that sustainable regulatory progress depends on open collaboration among agencies and industry. --- ### Harmonization Across Regions (≈ 00:01–00:05) > “How do you view harmonization efforts across regions, and can they truly align?” Panelists noted that harmonization does not require identical guidance across all regions. It rests on shared principles, not uniform language. They agreed that engagement, trust, and continued communication among regulators remain the enablers of practical harmonization. --- ### AI Guidance Philosophy & Risk-Based Approaches (≈ 00:19–00:32) > “We hear a lot about the phrase ‘take a risk-based approach’ when integrating AI. From a regulator’s perspective, what might this mean in practical terms?” — *Sarah Pope Miksinski, PhD* **Kevin O’Donnell** responded that most AI discussions start with technical capability recipes, processes, and data, but that’s only part of the picture. In regulation, the goal is to understand how much formality, transparency, and oversight are appropriate given the level of uncertainty and complexity. He explained that risk-based thinking must account for the program’s scale, the number of actors involved, and the intended use of the AI system whether for visual inspection, classification, or another regulated task. > “It’s more than just a risk assessment—it’s about formality, transparency, and oversight aligned to impact and uncertainty.” — *Kevin O’Donnell (HPRA)* He noted that this requires evaluating how human oversight, validation, and communication adapt to different AI architectures, especially when third parties or cloud systems contribute to the model’s operation. Roger Nosal agreed, emphasizing that risk assessments are valuable only when they connect directly to product efficacy, safety, and quality. Too often, he said, organizations limit their risk analyses to the next process step, missing the broader link to patient outcomes. > “Risk assessments must tie all the way to efficacy, safety, and product quality not just the next step in the process.” — *Roger Nosal (Vaxcyte)* The exchange reframed “risk-based” as a living framework, not a form to complete a method that integrates AI’s uncertainty, human oversight, and product impact into one continuum. **Tina Kiang** added that regulators are cautiously optimistic about AI’s efficiency potential, noting that technologies like digital twins can accelerate development and streamline manual processes when implemented responsibly. > “There are areas where perhaps using AI can make development faster… using technology to make the work more efficient.” — *Tina Kiang (FDA)* ### Forums and Legislation (≈ 00:33–00:41) > “Are there forums that bring industry, regulators, and lawmakers together to discuss AI and policy alignment?” — *Sarah* Panelists cited formal FDA listening sessions, international workshops, and Congressional communication channels as current venues. They also acknowledged tension between horizontal legislation (AI, data privacy) and vertical obligations (GxP, quality). Regarding the EU AI Act, several speakers suggested that requirements will depend heavily on context of use. They cautioned that legislation can quickly become outdated, while regulatory guidance allows faster iteration. --- ### Transition from Legacy to Digital Systems (≈ 00:45) > “What challenges should organizations anticipate as they transition from legacy to digital systems?” Panelists emphasized the goal of achieving structured submissions and improved knowledge management to foster regulatory reliance and reduce duplicate review. This vision aligns with the concept of a global quality dossier. **Tina Kiang** noted that many procedural requirements, like physical signatures, lag behind digital realities. Ongoing dialogue is needed to ensure laws and policies evolve with technology. Several speakers also raised concerns about a transparency gap, where agencies may use AI and internal data sources not visible to industry. They encouraged pre-competitive collaboration, publication of validation on methods, and shared learning forums to bridge this divide. This proved to be both an interesting point and something not exactly agreed upon by the entire room, which may really explain the fact of the matter: some believe the tech should be regulated, others want to experiment, while some are simply unsure where this jurisdiction will fall and how. --- ### **Summary Insight:** The Regulatory Town Hall reflected a confident and pragmatic tone: harmonization built on shared principles, AI discussed through a lens of quality and safety, and growing industry readiness to adopt digital systems. The underlying message, collaboration at pace, defined the day’s discussion. --- For questions, details, or inquiries, please reach out to hello@globalkeysolutions.net.