Date: 1–2 March 2027
Location: London, United Kingdom
Type: Summit
Website: securitysummit.ai
What Is the Global AI Security Summit 2027?
The Global AI Security Summit 2027 (GASS’27) is a two-day event taking place in London on 1–2 March 2027, built around a single organising principle: AI security is the entire programme, not a breakout track tucked into a broader conference. That focus distinguishes GASS’27 from the growing number of general AI events where security appears as one theme among many. Here, every session is dedicated to the question of how AI systems are secured in real-world settings.
The summit is explicitly aimed at the people who build, deploy, regulate and attack AI systems — a four-part audience that reflects the reality that AI security is not solely an engineering problem. Defenders, policymakers and offensive researchers all shape how the field evolves, and GASS’27 is structured to put those perspectives in the same rooms rather than in parallel silos.
Details on the organising body and programme committee are to be announced. What is clear from the published description is the event’s scope: production risk, securing models in critical infrastructure, and defensive automation form the core of the agenda. These are not abstract research topics but operational concerns for organisations running AI in live environments, which positions GASS’27 as a practitioner-oriented summit rather than a purely academic gathering.
Why It Matters for AI Professionals
AI security has moved from a specialist concern to a mainstream operational requirement. As models are embedded in critical infrastructure — the summit’s own framing — the consequences of failure, misuse or adversarial attack extend well beyond reputational damage. GASS’27 addresses this shift directly by concentrating its entire programme on production risk and defensive practice, giving attendees a dedicated forum rather than a side session at a general conference.
For professionals working in the field, the value of a single-focus summit lies in depth. Two days devoted entirely to AI security means more sessions on model protection, defensive automation and deployment risk than a multi-track event can offer. The summit’s inclusion of those who attack AI systems alongside those who build and regulate them also signals an emphasis on adversarial realism — understanding threats as they actually manifest, not as they are theorised.
What to Expect
The published programme description identifies three central areas of focus for GASS’27:
- Production risk: Sessions addressing the security challenges that arise when AI systems move from development into live production environments.
- Securing models in critical infrastructure: Content on protecting AI models deployed in infrastructure where security failures carry significant consequences.
- Defensive automation: Discussion of automated approaches to defending AI systems, reflecting the growing role of machine-driven security operations.
The summit’s framing around builders, deployers, regulators and attackers suggests the programme will span engineering practice, governance and offensive research. Specific speakers, session formats and a detailed agenda are to be announced; prospective attendees should check the official website for updates as the March 2027 dates approach.
Who Should Attend
GASS’27 is designed for a cross-functional audience rather than a single job function. Based on the summit’s stated aims, the core attendees include:
- AI and ML engineers responsible for building and deploying models in production.
- Security professionals — including red teamers and offensive researchers — working on adversarial threats to AI systems.
- Infrastructure and operations leaders securing AI deployments in critical environments.
- Regulators and policy professionals shaping the rules governing AI security.
- Technical executives accountable for AI risk across their organisations.
The common thread is practical involvement with AI security: people who need to understand how models are attacked, defended, deployed and governed, and who benefit from a programme with no competing tracks diluting the focus.
How to Register
Registration is handled through the official summit website at https://www.securitysummit.ai. Pricing, ticket categories and registration deadlines are to be announced — check the site directly for the latest information. Given the event’s focused scope and the two-day format in London on 1–2 March 2027, early attention to registration details is advisable once they are published.



