A new article by The Intercept details how the the agency can now automatically recognize spoken words by generating rough transcripts and phonetic representations that are easily stored and combed for information. The top-secret documents show NSA analysts congratulated themselves on developing what they called “Google for Voice” nearly a decade ago. It remains unclear how widely the spy agency uses its speech-to-text capabilities to transcribe and index U.S. citizens’ verbal conversations. The documents suggest the NSA has frequently used the technology to intercept phone calls — particularly in Iraq and Afghanistan, as well as Mexico — and to monitor international news. We are joined by Dan Froomkin, staff reporter at The Intercept.
JUAN GONZÁLEZ: “The Computers Are Listening.” That’s the headline to a new article by The Intercept detailing how the National Security Agency, or NSA, is converting people’s private phone conversations into searchable text. According to documents released by NSA whistleblower Edward Snowden, the agency can now automatically recognize spoken words by generating rough transcripts and phonetic representations that are easily stored and combed for information. The top-secret documents show NSA analysts congratulated themselves on developing what they called “Google for Voice” nearly a decade ago. It remains unclear how widely the spy agency uses its speech-to-text capabilities to transcribe and index U.S. citizens’ verbal conversations. The documents suggest the NSA has frequently used the technology to intercept phone calls, particularly in Iraq and Afghanistan as well as Mexico, and to monitor international news.
AMY GOODMAN: Well, for more, we go to Washington, D.C., where we’re joined by Dan Froomkin, a staff reporter at The Intercept. His new piece is called “The Computers Are Listening: How the NSA Converts Spoken Words into Searchable Text.” He also just wrote an article headlined “Why the USA FREEDOM Act is Both Desperately Important and Laughably Pathetic.”
Dan, welcome to Democracy Now! Talk about how voice is turned into text, and the significance of this for the NSA and for the public.
DAN FROOMKIN: Well, I mean, the funny thing is that it’s not at all surprising, when you think about it. I mean, we talk to our apps these days. We dictate into Google searches. So, is it—you know, is it any surprise the NSA can do this and has been able to do this for a while? No. But it has never been disclosed before, and the NSA refuses to talk about it. And so, because they don’t acknowledge that they do it, we can’t have a public policy discussion about what the limits to it should be, what the significance of it is. And the significance is enormous. Voice has historically been considered ephemeral and unsearchable. And, sure, if they—you know, if they had somebody around to listen to it, you might be spied on, if you were making international phone calls or what have you. But now they can listen to everything all at once, thanks to these computers. They can search through for whatever they might be looking for.
JUAN GONZÁLEZ: And does this extend also to, for instance, Internet phone calls or Skype, as well?
DAN FROOMKIN: From what I can tell, the NSA doesn’t really distinguish between any of these. They just call it voice intercepts. And they seem to be equally adept at pulling voice off of the standard old phone lines, phone circuits that travel across the fiber-optic lines, now packetized, you know, digital voice, so voice over Internet, Skype, whatever, cellphone conversations. Whatever it is, they pull it off, it’s voice intercept, and then—and then, at least theoretically, they could use this on any event.
AMY GOODMAN: Will the USA FREEDOM Act, the surveillance reform bill Congress is talking about, address this, and what you think of that?
DAN FROOMKIN: Well, it doesn’t address this at all. Of course, nothing addresses this, because nobody really knew about it before. But the USA FREEDOM Act, ironically, only touches on the one program, which we know doesn’t involve actual voice content, and therefore doesn’t involve voice recognition. It would just constrain the bulk collection of domestic metadata. I say “just.” That’s still a very significant move, but compared to the incredibly vast, sort of metastasizing surveillance state that has come to exist here, it’s really very little.
JUAN GONZÁLEZ: And for those listening or viewing the show who believe that their voice conversations should be kept private, what remedies are available to folks who don’t want to have their stuff grabbed this way?
DAN FROOMKIN: Well, at the end of the day, the answer is the same that we’ve heard to a lot of the Snowden revelations, and that’s encryption, really solid, unbreakable encryption. And there are—you know, Apple obviously is now doing that with its own iPhones and FaceTime, and you have services like Silent Circle and what have you. So, encryption would be the way to do it.
AMY GOODMAN: Finally, media monitoring by the U.S. government, here and abroad?
DAN FROOMKIN: Yeah, that actually struck me as one of the most anodyne uses of it, and actually, you know, perfectly fine, as far as I’m concerned. I’m happy the NSA wants to know what’s going on around the world, you know, collecting open source data. I don’t mind it. I’d like them to collect as much open source data as they can to find out what’s going on. The issue is when they’re looking at stuff that we think is private. So, if they’re translating an Arabic news channel into English and sending it out to their people, what’s wrong with that? You know, more power to them. But if they’re listening in on bedroom talk between the prime minister and his confidant, then that’s not so great—or maybe it is.
AMY GOODMAN: Dan Froomkin, I want to thank you for being with us, staff reporter atThe Intercept. We’ll link to your piece, “The Computers Are Listening.” And forDemocracy Now! viewers and listeners, you can watch and listen to our show online. For the NSA, it’s already transcribed, so you can read the whole thing there.