2) NSA likes to use "domestic" as a weasel word here for a number of reasons. The reality is that due to the FISA Amendments Act and its section 702 authorities, Americans’ communications are collected and viewed on a daily basis on the certification of an analyst rather than a warrant. They excuse this as "incidental" collection, but at the end of the day, someone at NSA still has the content of your communications. Even in the event of "warranted" intercept, it's important to understand the intelligence community doesn't always deal with what you would consider a "real" warrant like a Police department would have to, the "warrant" is more of a templated form they fill out and send to a reliable judge with a rubber stamp.
Glenn Greenwald follow up: When you say "someone at NSA still has the content of your communications" - what do you mean? Do you mean they have a record of it, or the actual content?
Both. If I target for example an email address, for example under FAA 702, and that email address sent something to you, Joe America, the analyst gets it. All of it. IPs, raw data, content, headers, attachments, everything. And it gets saved for a very long time - and can be extended further with waivers rather than warrants.
Their denials went through several revisions as it become more and more clear they were misleading and included identical, specific language across companies. As a result of these disclosures and the clout of these companies, we're finally beginning to see more transparency and better details about these programs for the first time since their inception.
They are legally compelled to comply and maintain their silence in regard to specifics of the program, but that does not comply them from ethical obligation. If for example Facebook, Google, Microsoft, and Apple refused to provide this cooperation with the Intelligence Community, what do you think the government would do? Shut them down?
Yes, I stand by it. US Persons do enjoy limited policy protections (and again, it's important to understand that policy protection is no protection - policy is a one-way ratchet that only loosens) and one very weak technical protection - a near-the-front-end filter at our ingestion points. The filter is constantly out of date, is set at what is euphemistically referred to as the "widest allowable aperture," and can be stripped out at any time. Even with the filter, US comms get ingested, and even more so as soon as they leave the border. Your protected communications shouldn't stop being protected communications just because of the IP they're tagged with.