JÜRI – Safe AI Use and Risk Mitigation

Target audience: AI users who want to gain practical skills for reducing risks, protecting their data, and checking the reliability of AI-generated results. This training continues the topics covered in the first workshop, but is also suitable for those who already have some experience using AI.

Understanding the risks of using AI is the first step. The next question is what we can do ourselves to reduce them. What information can we give to AI? How can we check its answers? How can we protect our documents and personal data? And how should we respond when AI, or a message created with its help, tries to mislead us?

The workshop focuses on practical techniques for using AI consciously and safely. We will look at how to phrase tasks more effectively, prepare information before sharing it, assess AI responses critically, and avoid dangerous or misleading recommendations. We will also explore how scammers use AI and how documents, websites, and other external materials may contain hidden instructions designed to influence AI.

The central idea of the training is simple: AI can be a highly capable assistant, but control and responsibility must remain with the human.

In the workshop, you will learn to:
- Develop safe AI-use habits and assess potential risks before starting a task.
- Prepare documents and data for working with AI, remove or conceal sensitive information, and understand the limitations of anonymisation and pseudonymisation.
- Use the privacy and security settings of AI applications and understand the difference between public, personal, and organisation-focused AI services.
- Check facts, calculations, sources, and conclusions in AI responses, using independent verification methods when needed.
- Recognise manipulative instructions, phishing attempts, and situations in which external text or a document tries to change AI’s behaviour.
- Create a simple set of safe AI-use guidelines for yourself to help decide which tasks to entrust to AI, what to check, and when to use other solutions.

In practical exercises, we will refine and make AI prompts safer, check suspicious responses, try out ways to protect data, and analyse situations in which seemingly harmless AI use can lead to unintended consequences.

Outcome: Participants will be able to use AI more thoughtfully and safely, apply practical techniques to protect data and check results, and develop simple risk-mitigation principles for their everyday use of AI.

How the workshop works
The workshop is practical and participant-centred, using demonstrations, group experiments, discussions, and short individual tasks. Participants are advised to bring a laptop and have access to at least one AI application (ChatGPT, Copilot, Gemini, Claude, or another). Exercises use prepared sample data so that participants’ personal or work-related information is not put at risk. You will receive sample tasks, checklists, and practical recommendations to help you develop safe AI-use practices.

The workshop is led by Henn Sarv – consultant, expert, practitioner, and trainer; IT Koolitus Trainer of the Year 2022; Microsoft Certified Trainer (MCT) since 1996; and Microsoft MVP in data science from 2004 to 2018.

The workshop is part of AI Week and the Eesti.ai initiative “The Most AI-Savvy Nation.”