No your code does not; but a poorly written client retrieving the results might subsequently try to fetch urls referenced in a generated DOCTYPE or xmlns attribute. It's bad clients making the requests based on a misunderstanding of the (correctly written according to the specs) document's contents.

Update: Just to clarify after rereading your post again: using CGI does not make any client calls in and of itself to retrieve anything from anywhere (w3c or otherwise). The HTML skeleton emitted by methods such as start_html includes http URIs which point to the w3c's servers which incorrectly written clients may attempt to retrieve after receiving the contents of your response.