LUChat: a secure generative AI tool for FGGA staff
FGGA is taking the lead in the implementation, goverance, and training of AI within Leiden University. This week, we take another step forward with the introduction of LUChat, a safe, secure, private generative Ai tool developed for use within Leiden University.
The reality we face
Every week, more than 1 million messages are exchanged with commercial AI tool providers such as ChatGPT, Claude and Gemini from Leiden University staff and students. This shows that generative AI is rapidly becoming part of everyday work. However, we do not have processing agreements in place with these companies, which means that the data entered into these tools is not protected under university safeguards. As a result, we risk exposing sensitive information and are not adequately protecting our research subjects, our intellectual property, or our confidential information.
In 2025, we introduced an AI Stoplight framework which tried to clarify how and when staff can safely process different types of information with commercial AI providers. While this already offers quite a lot of opportunity to work with Generative AI, the reality is that a lot of our most important work is restricted by either privacy, copyright, confidentiality, or intellectual property concerns. In other words, most of our work is in the red category in the stoplight model.
To overcome this, we need a technical solution. In collaboration with the ISSC, FGGA has taken an important step in introducing LUChat. From 30 June, all FGGA staff (including, of course, expertise centers and the LUC) will have access to LUChat through their ULCN account.
What is LUChat
LUChat offers an interface that will feel familiar to anyone who has used commercial AI tools, but with important differences behind the scenes. The tool runs within Leiden University’s secure cloud environment, users log in with their ULCN account, and prompts and outputs are not used to train public AI models.
The introduction of LUChat is part of FGGA’s 2026 and beyond faculty strategy. “FGGAi” is a broader programme for safe, responsible and practical use of AI. Over the past four months, the faculty has been piloting and testing the tool during its development phase, with a group of 40 staff in the so-called AI Pioneers Group. It has now reached a level of stability and maturity that makes it suitable for wider daily use by staff.
LUChat can support many common tasks, such as drafting or improving texts, summarising non-sensitive documents, brainstorming ideas, preparing communication, structuring notes, and exploring new topics.
Designed for privacy, transparency, efficiency
The tool is designed to run various different LLM models, giving the user control over whether they want to use a high energy or low energy model to process their request. The tool also gives users an overview of their usage in terms of tokens consumed and energy consumed. LUChat also implements RAG, meaning that chats are very economical in their context window. This makes the models more energy efficient.
Before using LUChat, staff are encouraged to read the user manual. The manual explains how the tool works, how data is handled, what the main functions are, and which types of use are appropriate.
Read the user manual here.
As with all AI tools, staff remain responsible for checking the output and for making sure that the tool is used appropriately. Not every type of data or task is suitable for generative AI. Existing rules around personal data, confidential information, copyright, research integrity and assessment continue to apply. For more information about this:
Access LUChat here.
Alongside the launch of LUChat, FGGA is also introducing an AI literacy module. AI literacy is essential for the careful and selective use of AI in our faculty. It starts with a shared understanding of how generative AI works, what it can and cannot do, and which legal, ethical and practical challenges it brings. The module builds on materials developed by LLInC and the University of Amsterdam and is already available to staff.
Access the AI literacy module here.
With LUChat and the AI literacy module, FGGA aims to give staff both the tools and the knowledge needed to experiment with AI safely, critically and constructively.
The official launch takes place this week. For questions, please contact Cameron Hope at c.a.hope@fgga.leidenuniv.nl.