The difference in response format, of JSON versus text, is expected in these different cases.
If the ID has the right format, the API will look it up, and return an error if it isn't found.
If the ID doesn't have the right format, e.g., it's the wrong length, it will fail validation, and the API won't bother trying to look it up.
You can check the response status code to see the different kinds of failures, and check the response Content-Type header to see what format the response body is in.
For example, using curl, these two calls look like: (responses trimmed for brevity)
curl -vX POST https://api.dropboxapi.com/2/users/get_account \
--header "Authorization: Bearer <ACCESS_TOKEN>" \
--header "Content-Type: application/json" \
--data "{\"account_id\": \"dbid:AACCsTU3d_-j5rfgJkGLObJErDLctQmUZw0\"}"
# < HTTP/1.1 409 Conflict
# ...
# < Content-Type: application/json
# ...
# {"error_summary": "no_account/.", "error": {".tag": "no_account"}}
curl -vX POST https://api.dropboxapi.com/2/users/get_account \
--header "Authorization: Bearer <ACCESS_TOKEN>" \
--header "Content-Type: application/json" \
--data "{\"account_id\": \"dbid:notvalidformat\"}"
# < HTTP/1.1 400 Bad Request
# ...
# < Content-Type: text/plain; charset=utf-8
# ...
# Error in call to API function "users/get_account": request body: account_id: 'dbid:notvalidformat' must be at least 40 characters, got 19
The API itself doesn't return the data/CDATA formatting in your post, so that is presumably coming from whatever tools you're using.
