We would like to thank the organizers for the kind invitation to contribute to this edition of Cutting Gardens. Silvia Storti presented the talk “Decoding brain signals with AI. What we can and cannot learn from EEG”. The talk discussed how AI can be used to decode EEG signals, but also why good performance is not enough on its own. When a model predicts something from EEG, we still need to ask what information it has used: the brain signal of interest, or another source of variance in the data. The theoretical session was followed by a hands-on lab led by Ilaria Siviero, where participants worked on EEG-AI examples and explored how different signal representations, models, and validation choices affect both performance and interpretation. We are grateful for the discussion and for the interest shown by the Cutting Gardens community. A special thank you to the local organizers from PandA Lab (https://panda.dsnm.univr.it/).