Lord Chris Holmes of Richmond MBE, Member, House of Lords — Chair, World AI Regulation Summit 2026
As artificial intelligence transitions from a technological race to a strategic geopolitical asset, global powers are aggressively defining their regulatory perimeters. This session examines how different governance philosophies—ranging from the EU's rights-based approach to China's state-controlled algorithmic sovereignty and the US/India models of innovation-focused regulation—are fracturing the global tech landscape.
Speaker: Gabriele Mazzini, Lead Architect and Drafter of the EU AI Act
Explore the fragmented U.S. AI policy landscape. This session unpacks the collision between rapid tech innovation and gridlocked federal legislation, state-level compliance patchworks, and the push for a unified national framework. Industry leaders will detail strategies to navigate current uncertainty.
Speaker: Alex Engler, Executive Director, Penn Center on Media, Technology, and Democracy
Where AI regulatory approaches converge and where they diverge across jurisdictions.
Moderated exchange between the EU and US keynote speakers, chaired to surface where the two frameworks converge, where they collide, and what that means for every other jurisdiction watching both.
Speakers: Lord Chris Holmes of Richmond MBE and Alex Engler
This session defines what countries and corporates must control to survive geopolitical and commercial shocks. We explore how nations secure infrastructure and jurisdiction, while corporates defend their competitive moats through proprietary workflow context and localized AI stacks.
What belongs at the model, deployment and infrastructure layers — and how to regulate frontier capability without freezing innovation. Widened to include how standards bodies (ISO, IEEE, ITU, OECD, NIST) are becoming the practical terrain for aligning infrastructure-layer rules across jurisdictions. This session unpacks responsibilities across the foundation model, deployment, and infrastructure layers. Industry leaders and policymakers will explore how to regulate frontier capabilities without stifling innovation, and how ISO, IEEE, ITU, OECD, and NIST frameworks are aligning global infrastructure rules.
Speaker: Gabriele Mazzini
Liability, evidence and human oversight when AI agents browse, decide, transact and escalate on their own. This session explores the critical challenges of liability, evidentiary tracking, and human oversight, equipping attendees with strategies to govern agentic systems effectively.
Speaker: Dr Mark Leiser, Professor, Leiden University Law School of Digital Regulation and Policy
Moderator: John Howell
| 13:30 – 14:10 | |
|---|---|
| TRACK A | TRACK B |
Update from the UAE / Saudi Arabia / GulfThe Gulf's sovereignty-first governance model — national AI strategy, cybersecurity oversight and regional data infrastructure. |
Update from Brazil / LATAMGlobal-South perspective on AI governance and public-policy priorities beyond the EU/US/China frame. |
| 14:10 – 14:50 | |
|---|---|
| TRACK A | TRACK B |
Update from SingaporeThe Model AI Governance Framework and AI Verify — operationalising testable, verifiable AI governance. |
Update from South KoreaImplementing the AI Basic Act — the first comprehensive AI statute enacted outside Europe. |
| 14:50 – 15:30 | |
|---|---|
| TRACK A | TRACK B |
Update from ChinaGenerative AI Measures, the Deep Synthesis Provisions and China's state-aligned compliance model. |
Update from South AfricaAn African governance perspective — not yet represented on the current roster. |
Court decisions in the US, UK and EU are setting precedent faster than legislatures can respond — training data, opt-outs, fair use and the emerging licensing market. This session analyzes cross-jurisdictional litigation across the US, UK, and EU, examining shifting precedents on fair use, the practical limits of data opt-outs, and the rapid evolution of licensing frameworks for generative AI training data.
Speakers: Bojana Bellamy, Alex Engler
Moderator: John Howell
A synthesis of the day's proceedings and the questions carried into Day 2.
Lord Chris Holmes of Richmond MBE
Dr Sepideh “Sepi” Chakaveh, Rapporteur, World AI Regulation Summit 2026
Day 2 opens on assurance, audit and compliance. The EU's high-risk obligations are biting before technical standards and supervisory tooling are ready — the first real test of whether risk-based regulation can be enforced in practice, and what other jurisdictions should take from it.
Keynote Speaker: Brando Benifei MEP, Member of the European Parliament; Co-Rapporteur of the EU AI Act
Widened to explicitly cover AI Assurance and Third-Party Testing — voluntary frameworks vs. mandatory audits, reframed around the live dispute between Anthropic's Advanced AI Framework (mandatory testing) and the White House's voluntary 30-day review order. To include country-specific briefers recycled from other sessions; panel of at least two, moderated by John.
Brando Benifei MEP
Dr Mark Leiser
Moderator: John Howell
| 10:30 – 11:30 | |
|---|---|
| TRACK A | TRACK B |
Financial ServicesModel risk management, algorithmic trading oversight and AI in credit/underwriting decisions under the EU AI Act's high-risk regime, UK regulator guidance (FCA/PRA/Bank of England) and US state-level rules. Conny Dorrestijn Moderator: John Howell |
HealthcareClinical decision-support AI, diagnostic tools and medical device classification — how regulators are drawing the line between a “wellness app” and a regulated medical device. |
| 11:30 – 12:30 | |
|---|---|
| TRACK A | TRACK B |
Pharmaceuticals and Life SciencesAI in drug discovery, clinical trial design and regulatory submissions — where existing pharma regulation already anticipates (or doesn't) AI-generated evidence. |
E-Government and Public SectorAutomated eligibility decisions, public-service chatbots and algorithmic transparency registers — accountability standards when the state itself is the deployer. |
Practitioner “how to” training sessions, each led by a lawyer or an activist.
| 13:30 – 14:00 | |
|---|---|
| TRACK A | TRACK B |
How to Build an AI Risk Register That Survives an AuditHow to establish a comprehensive AI use-case inventory, align with recognized governance frameworks (such as ISO/IEC 42001 or NIST AI RMF), and maintain auditable evidence like decision records, control checklists, and testing results. Dr Mark Leiser |
How to Run a Public-Interest Algorithmic AuditRunning a public-interest algorithmic audit involves systematically evaluating an AI or machine learning system for biases, harms, and transparency. Because platforms fiercely protect their source code, public-interest audits typically rely on external "black-box" testing methodologies, such as creating controlled experimental environments, querying APIs, or deploying audit studies to measure real-world outputs. Reema Patel Dr Jeni Tennison |
| 14:00 – 14:30 | |
|---|---|
| TRACK A | TRACK B |
How to Conduct a Conformity Assessment Under the EU AI ActConducting a conformity assessment under the EU AI Act requires developers of High-Risk AI Systems to evaluate compliance across key areas: risk management, data governance, technical documentation, transparency, human oversight, and cybersecurity. Providers must do this before placing their systems on the market. Gabriele Mazzini, Lead Architect and Drafter of the EU AI Act |
How to Build Community Voice into Data and AI GovernanceBuilding community voice into data and AI governance shifts the process from a theoretical top-down exercise to a democratic, transparent practice. It involves institutionalizing feedback loops, ensuring diverse stakeholder representation, and co-designing policies directly with the populations most affected by automated decision-making. Jeni Tennison, Executive Director, Connected by Data |
| 14:30 – 15:00 | |
|---|---|
| TRACK A | TRACK B |
Cross-Border AI Compliance — What Organisations Must Prepare ForOrganizations must inventory all AI exposure, implement privacy-preserving technologies (like differential privacy), and establish cross-functional governance frameworks to prevent severe financial and reputational penalties. |
How to Campaign for AI Legislation: Lessons from the FieldCampaigning for AI legislation requires building broad coalitions, mobilizing grassroots support, and engaging directly with lawmakers using clear, evidence-based policy proposals. By translating complex technical risks into relatable human impacts, advocates can successfully push for responsible artificial intelligence governance. |
Export controls on chips and frontier models, autonomous-systems governance, and the cyber/bio dual-use risks industry leaders raised at the 2026 G7 — the sovereignty question made concrete.
Sovereign AI empowers nations to develop, own, and govern AI models using locally hosted infrastructure and data. Driven by stringent cross-border data restrictions, global AI regulation is increasingly fractured. Organizations must now adopt decentralized architectures to comply with localized regimes.
Lord Chris Holmes of Richmond MBE
Emmanuel Daniel
Brando Benifei MEP
Bojana Bellamy
Moderator: John Howell
*The agenda is subject to change. Content, speakers, and timing will be updated periodically. Please check back regularly for the latest updated agenda.