GDPR News » IAPP AI Governance Global North America 2025

Key takeaways from the panels:

• 🔑 Responsible data curation lies at the heart of trustworthy AI. Managing data transparently across its lifecycle, from collection and labeling to cleaning and ongoing maintenance, is crucial to ensure fairness, accountability, workers’ rights, and respect for human rights.

• 🌍 The geopolitical AI race is reshaping the landscape. The US is showing more policy continuity than expected, even as administrations change. While Beijing moves quickly on decision-making, the US leverages creativity and innovation, though energy and infrastructure challenges remain major friction points.

• 🏥 Privacy Enhancing Technologies (PETs) are creating new opportunities for responsible health data use. Synthetic data, federated learning, and evolving interpretations of anonymisation and pseudonymisation were highlighted as ways to balance innovation with privacy.

• 📊 AI impact assessments are gaining traction globally, but adoption is uneven. Many companies are integrating them into existing risk management processes, though measuring effectiveness and getting full transparency from providers remain significant challenges.

• 🤝 Government and industry are finding common ground. Transparency should focus on what is actually useful for end users, models must be secured against attacks, and meaningful human oversight is critical when AI systems can affect lives. Deepfakes and bias remain urgent areas where proactive regulation is needed.

• 🔍 Explainability is practical, not just theoretical. Global explainability helps regulators and operations teams assess model performance, while local explainability supports clinicians and compliance teams in evaluating individual outputs. Documentation, oversight, and training are essential to make explanations meaningful.

• ⚖️ Legal and policy debates on copyright and training data are intensifying. We attended an excellent session under the Chatham House Rule that raised important questions about copyright and AI training, but details cannot be shared.
  

• 📚 Navigating global AI regulation requires clarity. A taxonomy comparing frameworks such as the EU AI Act, US Executive Order, and China’s interim measures helped cut through fragmented definitions, highlighted common ground, and exposed gaps, moving the conversation beyond the false dichotomy of innovation versus regulation.

• ✈️ We were happy to attend and look forward to the next conferences. EDPO will now fly back to Brussels to attend more events in Europe, and next month we will return across the Atlantic for the IAPP conference in San Diego on 30–31 October.