Luxembourg releases open-source sandbox for trusted AI testing

Luxembourg has released the AI Assessment Sandbox Configurator, an open-source tool for building customised environments to test whether AI systems are trustworthy and compliant. Launched in June 2026 by the Luxembourg AI Factory, the tool was co-developed by the Luxembourg Institute of Science and Technology and the University of Luxembourg’s SnT.

The Configurator draws on a catalogue of AI Tests & Controls, an extensible library of testing tools, controls, and datasets built through an open plug-in API. Organisations can select the elements relevant to their use case, then generate a sandbox with test execution, a collaborative dashboard, report generation, and tamper-proof logs for audits.

The system is designed to reduce the manual work involved in choosing tests, setting metrics, preparing sandboxes, and aligning technical, legal, and business teams. It is intended to support AI deployments across sectors including banking, manufacturing SMEs, public administration, and startups, while helping organisations address obligations created by the EU AI Act.

LIST’s related AI Sandbox work has already been used with Banque Internationale à Luxembourg to assess banking chatbots, with the City of Luxembourg on public-service AI applications, and with Mistral AI. The Configurator, its catalogue, and source code are available on GitHub, with organisations invited to use it, extend the catalogue, or contribute to the codebase.