added disclaimer

app.py

@@ -548,6 +548,8 @@ with block as demo:
             <b>How does it work?</b> We are giving CLIP your images and let it choose from 1000 possible names. 
             As CLIP is predicting the names that match the given images, we can probe whether the model has seen your images 
             during training. The more images you upload the more confident you can be in the result!
+            <br><br>
+            <b>Disclaimer:</b> In order to process the images, they are cached on the server. The images are only used for predicting whether the person was in the training data.
           </p>
         </div>
         """