License under CC BY-NC 4.0; add Intended Use / Out-of-scope / Risks section

LICENSE

@@ -0,0 +1,12 @@
+NEvo is licensed under the Creative Commons Attribution-NonCommercial 4.0
+International License (CC BY-NC 4.0).
+
+You are free to share and adapt the material for non-commercial purposes,
+provided you give appropriate credit. Commercial use is not permitted under
+this license.
+
+Human-readable summary: https://creativecommons.org/licenses/by-nc/4.0/
+Full legal code:        https://creativecommons.org/licenses/by-nc/4.0/legalcode
+
+The referenced upstream models (vJEPA-2, SDXL-Turbo, LTX-Video) are covered by
+their own separate licenses and access terms.

README.md

@@ -1,4 +1,5 @@
 ---
+license: cc-by-nc-4.0
 tags:
 - neuroscience
 - fmri
@@ -213,3 +214,14 @@ Project website: [nevo-project.epfl.ch](https://nevo-project.epfl.ch/)
 ## Acknowledgements
 
 Builds on BrainDiVE-style encoder-guided synthesis, vJEPA-2, SDXL-Turbo, and LTX-Video. ROI/searchlight definitions derive from an fsaverage-space group atlas (precomputed and bundled).
+
+
+## Intended Use, Out-of-Scope Use & Risks
+
+**Intended use.** NEvo is a **research-only** tool for computational visual neuroscience. It searches for image/video stimuli that a differentiable image/video→fMRI *encoder model* predicts would drive a chosen brain region or target fMRI pattern. Its outputs are computational **hypotheses** — candidate stimuli to be validated in properly designed, ethically approved experiments — not established scientific findings.
+
+**Out-of-scope use.** This model is **not a medical device** and must not be used for diagnosis, treatment, or any clinical decision-making. It is **not intended for, and must not be used to design, guide, or deliver, any form of real brain stimulation** (e.g. TMS, tDCS/tACS, deep-brain or intracranial stimulation) or any other in-vivo neuromodulation. It does **not** model the effect of physical stimulation on a living brain — it only optimizes against a fixed encoder's *predicted* fMRI response. Any use on, or affecting, human or animal subjects requires independent ethical review (e.g. IRB/ethics-board approval) and is the sole responsibility of the user.
+
+**Risks & limitations.** Optimizing hard against a single encoder can exploit encoder artifacts, so generated stimuli may not transfer to real neural responses — treat every output as an unvalidated hypothesis and use held-out validation. The underlying generative models can produce biased, unexpected, or disturbing imagery. The authors provide the model **as is**, without warranty of fitness for any purpose, and accept no liability for downstream use.
+
+**License.** Released under **CC BY-NC 4.0** (Creative Commons Attribution-NonCommercial 4.0): research and other **non-commercial** use only, with attribution. The referenced upstream models (vJEPA-2, SDXL-Turbo, LTX-Video) carry their own licenses and access terms, which you must also comply with.