Safe Artificial Intelligence Lab

SAIL is a research group at Imperial College London led by Prof. Alessio Lomuscio. We develop methods and tools that make AI systems safe to deploy. Applications range from perception and sequential decision-making in autonomous systems to conversational assistants and agents running business workflows. We find their vulnerabilities, establish formal guarantees on their behaviour, train them to be robust, and monitor them after deployment with a focus on making them explainable.

We work with academic and industrial partners in the UK, Europe and the US, guided by the belief that AI should be safe and secure for society to use.

The SAIL team, November 2025
The lab, November 2025.

Current research areas

The lines of research that current members of the lab are developing.

All research areas →

Software

Verification tools developed in the lab, released under open-source licences.

VenusVerification of ReLU networks combining mixed-integer programming, dependency analysis, bound propagation and splitting.GitHubPaper
VeriNetVerification based on symbolic interval propagation, adaptive refinement and adversarial search.GitHubPaper
MCMASSymbolic model checker for multi-agent systems against temporal, epistemic and strategic specifications.DownloadManual

All software →

SAIL Seminar

Join SAIL

We take new PhD students every year, including through the UKRI Centre for Doctoral Training in Safe and Trusted AI. Postdoctoral positions are advertised here when available.

PartnershipsDARPA Assured Autonomy programUKRI Centre for Doctoral Training in Safe and Trusted AIRoyal Academy of Engineering