I'm not sure you understand the question. It's not a matter of IP addresses. Protected tweets are not publicly accessible, so it's not a given that intelligence agencies have access to them unless Twitter is providing it to them, or they've hacked into them.
Twitter's privacy policy doesn't actually require a warrant to give up your "private" info (including your protected tweets).
Law and Harm: We may preserve or disclose your information if we believe that it is reasonably necessary to comply with a law, regulation or legal request; to protect the safety of any person; to address fraud, security or technical issues; or to protect Twitter's rights or property.
I doubt Twitter gives them everything since it's a difficult engineering problem to handle that much data. But I could easily believe the FBI has a list of terms or phrases that Twitter self-monitors for.
