Luxembourg AI Factory offers AI Providers and AI Deployers the opportunity to assess the trustworthiness of their AI systems using an innovative open-source tool. 

Luxembourg AI Factory aims to help organisations adopt AI in a responsible way. This includes performing testing of AI solutions to ensure that they can be fully trusted by employees, customers and other relevant stakeholders, thus opening the way to wide adoption and value creation.  

Testing the trustworthiness of AI solutions can be very time consuming and requires specialised skills that are not yet available in most organisations, especially SMEs and start-ups. That is why the Luxembourg AI Factory is offering support with experts and an innovative open-source tool: the AI Assessment Sandbox Configurator, developed by Luxembourg Institute of Science and Technology and the University of Luxembourg’s SnT. This tool significantly reduces the time and cost of assessing AI systems.significantly reduces the time and cost of assessing AI systems. 

What you receive is support to set up an AI technical sandbox for self-assessment with the assistance of independent technical experts from Luxembourg AI Factory.  

It is important to note that the Technical Sandbox pilot is not a Regulatory Sandbox under the supervision of a Competent Authority (AI Act Art. 57 and 58), nor is it a conformity assessment. Finally, the Technical Sandbox pilot does not lead to certification. 

• Organisations that have a legal entity registered in Luxembourg, or a branch or permanent establishment in Luxembourg 
• Organisations that are developer, provider, deployer or prospective deployer of an AI system
• Organisations that want to assess an AI system at the stage of having a working solution that is mature enough to be tested, not a concept or a design on paper
• Organisations that can provide access to the AI system to run the tests. You do not need to open up your system, black-box testing works too.
• Organisations that can commit time of their team to perform the assessment, typically technical experts as well as business, risk management and legal experts. The effort will range between 3 and 20 working days over an elapsed period between 6 and 12 weeks. More detailed effort and elapsed estimates will be defined after the first workshop.
• Where relevant, organisations that can provide data and infrastructure required to run the tests. More details on required data and infrastructure will be defined after the first workshop.
• While we strive to accommodate all requests, the Luxembourg AI Factory may reject applications that do not meet the above-mentioned requirements.

We encourage interested stakeholders to fill out this application form. Luxembourg AI Factory is available if you have any questions or need support to fill out the application. 

Once you submit the application form, you will be invited to a discovery workshop via web meeting or in person. At the end of this workshop, you will receive a draft AI system assessment plan with details on the types of tests to be performed, estimate of duration, effort, data and infrastructure required, etc. Once you sign-off the plan, there will be a technical workshop to set up the AI technical sandbox. Then the planned test will be performed.  

Based on the test results, one or more iterations might be required. At the end of the assessment, you will receive a report from our team summarizing the tests performed, the findings and the suggested actions. Finally, you will be asked to fill out a feedback form.