Agenda

►AI in the Middle: Basic security considerations in multi-model/agentic architectures

Ben Trethowan, CISO, Brit Insurance

• Identifying the need to perform effective supply chain security assessments upon AI providers, including consideration of product/service-level aspects.
• Discussing the importance of effective Non-Human Identity (NHI) management, across user/workload identities, as the role of identity encapsulation/impersonation increases.
• Highlighting the challenges of achieving pervasive security governance and observability across complex AI architectures.
• Appreciating the potential ‘novel’ security threats introduced through AI utilisation.
• Leveraging existing data security approaches to support the extension of data visibility/control into complex AI architectures.

►Can You Trust the AI You Buy?
Rethinking procurement, assurance and accountability in the age of AI

Simon Brady, Event Chairman (AKJ Associates)
Leroy Stanford, Director of Cyber Security, Ocado Group
Sarah Lawson, Director of Cybersecurity, Risk and Resilience (CISO), Oxford University Press

• Rethinking procurement, assurance and accountability in the age of AI
• How has the arrival of AI changed the way your organisation assesses and buys technology?
• Have you changed where accountability sits when approving technology that incorporates AI — and who now needs to be involved?
• Are your third-party contracts keeping pace with AI — particularly around data, security, transparency, assurance and liability?
• Can we ever have sufficient visibility into the AI embedded across our technology supply chains — and if not, how do we manage the risks we cannot see?

►Signal in the Noise: AI, Hype, and the Risk CISOs Should Actually Worry About

Andy Giles, Executive Director, Cyber & Technology Risk Reporting and Metrics, JPMorgan Chase

• Cut through the AI hype to understand the real risks already emerging in business, using real-world incidents and current data.
• Explore why the biggest threat is not just malicious AI, but the speed of AI adoption outpacing governance and security.
• Gain practical, no-jargon guidance on enabling AI safely, managing it as a business risk, and leading the conversation with confidence.

►Outside the US and China, can ‘the rest’ really secure AI?

Isabella Wilkinson, Senior Research Fellow, Digital Society Programme, Chatham House

• Far from the front of the AI race, many countries are worried about their extreme dependence on US and Chinese technology, and the security risks generated by the breakneck pace of AI development and diffusion
• Without superpower-level access to compute and capital, and unsteady access to frontier capabilities, can ‘the rest’ secure the AI that increasingly underpins their societies?
• What can ‘the rest’ do together on AI security that they can’t do apart?

►Data Without Borders: How Far Can Your Data Travel
Rethinking control, trust and exposure in the age of AI

Simon Brady, Event Chairman (AKJ Associates)
Andy Powell, CISO, A.P. Moller Maersk
David Varney, Partner, Womble Bond Dickinson
Ray Dodd, Head of Infosec Transformation & AI Security, Just Group PLC

• What does it mean to truly control your organisation’s data in an AI-driven world?
• As organisations embrace external AI platforms and services, how much visibility and control over their data are they prepared to give up?
• Does bringing AI in-house fundamentally change the risk — or simply shift where responsibility and accountability sit?
• As AI becomes embedded across the enterprise, who should ultimately be accountable for how organisational data is accessed, used and protected?

►The Deadline Moved. The Declaration Didn't.

Punit Bafna, Head of Cybersecurity & Compliance, Haleon

• What actually changed under the AI Act amendments in July 2026, and which obligations were never deferred at all
• Why standard identity controls (recertification, joiner-mover-leaver, segregation of duties) fail structurally when the identity is an autonomous agent
• Where the current control frameworks stop short, and what evidence a board realistically needs before declaring controls effective
   

►When "Vibe Coding" Collides with Production Reality: The Enterprise Security Bottleneck

Ajay Kumar, Head of AI, Salesforce

• Where AI-generated code collides with enterprise security architecture
• How AI-assisted development changes the application security threat model
• How security can evolve without becoming the bottleneck

►Stop Misbehaving! Why AI agents don't follow their own rules

Etay Maor, VP Threat Intelligence and Co-Founder of Cato CTRL, Cato Networks

• We read about AI agents performing malicious activities that they were not designed, and sometimes explicitly forbidden, to do.
• This raises two questions: why do they do this, and how can they do this if there are guardrails.
• In this session we will explore the different ways and reasons AI agents lose site of their true objectives, how threat actors can use this in their advantage, and what strategies and approaches organizations can take to detect, mitigate, and prevent this behavior.
• Stop misbehaving, bad agent!

►Ask Again Next Quarter: Turning the UK’s AI Risk Toolkit into a Functional ISO 42001 Program

Patrick Sullivan, Vice President of Strategy & Innovation, A-LIGN

• Understand how to use the UK DSIT AI risk toolkit gives you (the right questions for each AI system) and what it leaves out: ownership, internal audit, and management review, at a time when 97% of organizations with an AI-related incident lacked proper AI access controls
• Learn practical strategies for running the toolkit inside an ISO 42001 management system, with a real-world example of an embedded SOC triage copilot from prompt-injection risk to a treated, owned, and reassessed result
• Leave with a 30/60/90-day plan to inventory embedded AI, run the toolkit on your top systems, and take the register to internal audit and management review

►AI Security Is a Data Security Problem: Why winning with AI depends on what you let it reach

Stuart Wilson, Sales Engineering Director, Europe, Forcepoint

• Understand why the data AI can reach, rather than the model you choose, determines both the value and the risk of AI.
• Learn how AI assistants, copilots and agents are changing the way sensitive data is accessed, created and shared.
• Recognise where traditional security approaches fall short as AI adoption outpaces policy.
• Discover what it takes to stay in control of AI whilst enabling the business to move at pace. 

►Assessing the Maturity of AI in Cyber Threat Operations

Aiden Sinnott, Principal Threat Researcher, Sophos

• Artificial intelligence has rapidly become embedded within the cyber threat landscape, but separating genuine operational capability from media hype remains a challenge.  
• How threat actors are adopting AI across the attack lifecycle, from phishing and malware development to reconnaissance, social engineering and emerging autonomous attack workflows 
• Where AI is delivering measurable advantages to attackers today, where human expertise remains indispensable, and how the next generation of agentic AI could reshape offensive cyber operations  
• The current state of AI-enabled threats, the trajectory of future developments, and the implications for cyber defence 

►From AI Adoption, to AI Assurance

Callum Ring, AI Innovation Architect, Bytes

Khetan Gajjar, Field CTO EMEA, Mimecast

•  AI adoption is accelerating, but security, governance and control need to move just as fast. 
• In this keynote, Bytes, Microsoft and Mimecast will explore what it takes to move beyond AI experimentation towards confident, secure and governed adoption.
• Bringing together strategic and market perspectives, the discussion will explore the challenges emerging as AI becomes embedded across the enterprise, from shadow AI and workforce behaviour to sensitive data moving through new tools and platforms. 
• The session will uncover how organisations can strengthen visibility, control and accountability, turning rapid AI adoption into a practical approach to AI assurance.