Yes the files will be visible through the web interface - but if there’s an ai application through which the users are interacting w/ the assistant - then the uploaded document is accessible to that user only (unless they somehow know what other documents have been uploaded and what their content is).
The files object is independent of the assistants and threads and that’s where the underlying issue is. One can associate files to assistants but not to threads. I think it should have been the other way around or both should have been possible (associate files with assistants and with threads. If its associated only with thread, then it’s like a session based file upload and some kind of garbage collector could clean that up once the thread is terminated. If its associated with assistant, then its persistent and available across all threads). What do you think?
Or How else would you suggest we handle this situation from application side? multiple people are interacting w/ the assistant and each wants to analyze their uploaded file. if the file doesn’t get associated to the assistant, it won’t be able to find it as everyone is reporting above.