The Turing Lectures: Frontier AI under pressure - building resilience across layers

The Alan Turing Institute’s flagship public lecture series returns for its 10th year, bringing world-leading thinking in data science and AI to the forefront of national conversation. In 2026, the series turns its focus to resilience. As rapid technological change reshapes our world, AI is transforming how we protect and strengthen the systems we depend on – from national infrastructure and defence to climate and environmental sustainability. While new capabilities offer powerful tools to anticipate risk, enhance preparedness and respond to complex global challenges, they also introduce fresh vulnerabilities and difficult trade-offs. In this year's lectures, we will explore how AI can bolster resilience across critical domains, while also addressing the societal implications of its deployment. Bringing together Turing researchers and invited experts, the lectures will provide rigorous insight and practical perspectives on the foundations of our shared future. About the event How do we make our institutions resilient to frontier AI? What does it really take? Frontier AI – highly advanced AI systems, such as GPT-5 or DeepMind’s Gemini, that push the limits of what machines can do – is moving faster than our institutions can keep up. This creates challenges that aren’t just technical, but affect the whole system, including AI models, policies, organisations and human decision-making. Join us for the first of our Turing Lectures to explore resilience not as a simple technical fix, but as a broader challenge that spans AI, institutions, policy and human judgement. Dr Mikel Rodriguez has led research within the Frontier & AGI Security Group at Google DeepMind and, as a Fellow at MITRE Labs, has built and led the Department of Defense’s AI Red Team as well as co-creating ATLAS – the globally adopted framework for adversarial threats to machine learning. In this Turing Lecture, Mikel draws on his hands-on experience in frontier AI research, national security and cyber operations to explore the pressure points where rapid AI advances meet fragile institutions – and what it really takes to build systems that can adapt. The answer doesn’t lie in perfect control, but in humility, diversity and the courage to act decisively amid uncertainty. 00:00 Welcome and Theme 00:52 Resilience and Institutions 03:04 Meet Dr Mikel 04:47 Turing to Dartmouth Origins 08:29 Early Failures and Winters 13:06 Deep Learning Breakthrough 15:44 Transformers and Scale 19:08 Self Improving AI Timelines 20:11 Institutions Lag Behind 21:59 Pressure Point 1 Cyber Dual Use 28:18 Pressure Point 2 Proliferation 35:25 Pressure Point 3 Counter AI 41:24 Pressure Point 4 Monoculture 44:18 Pressure Point 5 Asymmetry 45:45 Pressure Point 6 Policy Velocity 46:54 Unifying Pattern Across Gaps 47:42 Defining Resilience Margin 49:27 Posture Under Uncertainty 50:41 Turing and Knowable Limits 51:54 Governing Thinking Machines 53:48 Speed Versus Capacity 56:42 Geopolitics and Regulation 59:45 Government Capacity Gaps 01:02:22 Data Diversity and Inclusion 01:06:26 Education and Cognition 01:11:18 Democracy and Institutions 01:14:54 Sentience and Self Improvement 01:18:55 Career Resilience Advice 01:22:09 Multi Agent System Risks 01:25:53 Closing Thanks and Wrap