AMLUCS 2026: UK’s Leading Government-Backed Conference on Applied AI and Cybersecurity Comes to London

The fourth annual Applied Machine Learning for Cyber Security (AMLUCS) conference is set to take place on 23–24 September 2026 at Queen Elizabeth Olympic Park in London, with a pre-conference training course on 21–22 September.

Positioned as the UK’s only annual, government-backed event dedicated to the practical intersection of artificial intelligence and cybersecurity, AMLUCS brings together security engineers, data scientists, researchers, vendors, policymakers and operators focused on real-world application rather than hype.

Organised as a not-for-profit, practitioner-led conference, AMLUCS has grown significantly for 2026.

It features a new Technical Committee, a twin-track format, a larger targeted audience of around 120–150 attendees, increased diversity across sectors, and deeper government partnerships.

The event emphasises applied research and operational practice in areas including AI security threats, vulnerabilities and mitigations; agent capabilities in offensive and defensive cyber operations; multi-agent systems; AI security governance, risk and standards; and explainability and fairness in security contexts.

Event Partners and Sponsors

AMLUCS 2026 is brought to you by event partners Frazer-Nash Consultancy, the Defence Science and Technology Laboratory (dstl) and the National Cyber Security Centre (NCSC), part of GCHQ.

Platinum sponsors are Microsoft and Abundant Security.

Gold sponsors include Actica Consulting, Deep Cyber and Tenacium DC.

The NCSC has publicly confirmed its partnership role, underscoring the conference’s government-backed credentials.

Programme Highlights

The highly curated programme includes keynotes, practitioner panels, a hands-on workshop, technical presentations and posters, plus the new two-day training course.

• Keynote: Hyrum Anderson, cofounder and CEO of Abundant Security (previously Senior Director of AI & Security at Cisco, CTO at Robust Intelligence, and roles at Microsoft, Mandiant and elsewhere). Anderson co-founded the Conference on Applied Machine Learning in Information Security (CAMLIS), a key inspiration for AMLUCS. A second keynote is to be announced.
• Panels: “Where Do We Start? A Practical Guide for Organisations Adopting AI-Driven Cybersecurity” and “Secure Enough? How Regulation Shapes AI Security.”
• Hands-on workshop: “The Agent’s Inner Monologue: Catching Compromise in the Reasoning Trace” (Secure Agentics), using the open-source Adrian platform to monitor agent reasoning for attacks missed by action-level monitoring.
• Training: A two-day course on validated AI red teaming methodology for production AI systems (aligned with ETSI EN 304 223), covering foundations, multi-turn techniques, methodology, reporting and conformity assessments.

Technical Presentations and Posters

The technical programme is organised into two main tracks reflecting the conference themes.

AI Security Threats, Vulnerabilities & Mitigations features work from organisations including:

• BAE Systems Digital Intelligence (Agentic Large Language Models for Autonomous Cyber Operations)
• Microsoft AI Safety Platform (Cross-Layer Detection for Scaled AI Misuse)
• University of Oxford (Detecting Attacks Against AI Systems)
• HiveTrace (Do Model Representations Know When a Jailbreak Is Working?)
• Alan Turing Institute (Hessian Spectral Signatures for Backdoor Detection)
• Loughborough University (Latent Backdoors)
• BMT (Packets as Binary Images: Per-Packet Intrusion Detection for OT)
• Omega (Per-Agent Compliant, Collectively Unsafe…)
• Deep Cyber Ltd (Planning for Quantum Backed AI)
• Palo Alto (Reproducible Agentic Environment System – RAES)
• DSTL (Security of Novel Compute in the AI Stack)
• Queen’s University Belfast (The Agentic Attack Surface: Latent Failures in Autonomous Workflows)
• Lancaster University (Thematic Review and Gap Analysis of AI Security)
• Alice (Two Graders for One Agent…)

Agent Capabilities, Offensive & Defensive Cyber Operations includes contributions from:

• The Alan Turing Institute & University of Cambridge (Addressing the Root Cause: Automated Fuzzing Harness Generator…)
• Advai (Agentic Security Benchmarks are Executable Threat Models…)
• Cardiff University (Auditing and Recovering Agentic Cyber Defence Systems)
• BT (Automated Experimentation in Cyber Labs)
• Alice (Capture the Flag, Miss the Malware…)
• Riskaware (From Detection to Response)
• Loughborough University (Identifiable Causal Discovery…)
• Deep Cyber Ltd (Investigating Agentic Cybercrime)
• Aeris-UK (Measured Feasibility of Agentic Cyber Defence at the Land Tactical Edge)
• Frazer-Nash Consultancy (Novel Compute for Data-Efficient Reinforcement Learning…)
• NCSC (Project AIX: Deploying Cyber Resilience Agents on a Real Network)
• Palo Alto (Sustainable Cybersecurity in the Era of Hyperconnected Agentic AI…)
• Tikos Technologies (Reasoning, Not Rationalising: Mechanistic Guardrails…)
• Vidoc Security Lab (Reproducing Mythos Findings… and Disclosing Two New Linux Kernel Zero-Days)

These sessions showcase cutting-edge research and practical deployments from government, industry and academia on agentic AI risks, backdoors, jailbreaks, autonomous cyber defence, collective agent safety, quantum considerations, and real-network resilience.