WASHINGTON — The National Security Agency (NSA) and the FBI are tapping directly into the central servers of nine leading U.S. Internet companies, extracting audio and video chats, photographs, emails, documents and connection logs that enable analysts to track one target or trace a whole network of associates, according to a top-secret document obtained by The Washington Post.
The program, code-named PRISM, has not been made public until now. It may be the first of its kind. The NSA prides itself on stealing secrets and breaking codes, and it is accustomed to corporate partnerships that help it divert data traffic or sidestep barriers. But there has never been a Google or Facebook before, and it is unlikely there are richer troves of intelligence than the ones in Silicon Valley.
Equally unusual is the way the NSA extracts what it wants, according to the document: “Collection directly from the servers of these U.S. Service Providers: Microsoft, Yahoo, Google, Facebook, PalTalk, AOL, Skype, YouTube, Apple.”
PRISM was launched from the ashes of President George W. Bush’s secret program of warrantless domestic surveillance in 2007, after news-media disclosures, lawsuits and the Foreign Intelligence Surveillance Act (FISA) Court forced the president to look for new authority.
- McMorris Rodgers should ask hometown folks about Obamacare
- Seattle congestion: We're No. 5
- Expedia expected to announce Seattle move
- Seahawks re-sign FB/DL Will Tukuafu
- Seattle traffic congestion: We're No. 5
Most Read Stories
Congress obliged with the Protect America Act in 2007 and the FISA Amendments Act of 2008, which immunized private companies that cooperated voluntarily with U.S. intelligence collection. PRISM recruited its first partner, Microsoft, and began six years of rapidly growing data collection beneath the surface of a roiling national debate on surveillance and privacy. Late last year, when critics in Congress sought changes in the FISA Amendments Act, the only lawmakers who knew about PRISM were bound by oaths of office to keep silent.
The court-approved program is focused on foreign communications traffic, which often flows through U.S. servers even when sent from one overseas location to another. Between 2004 and 2007, Bush administration lawyers persuaded federal FISA judges to issue surveillance orders in a fundamentally new form. Until then the government had to show probable cause that a particular “target” and “facility” were connected to terrorism or espionage.
In four new orders the court defined massive data sets as “facilities” and agreed to occasionally certify that the government had reasonable procedures in place to minimize collection of “U.S. persons” data without a warrant.
Several companies contacted said they had no knowledge of the program and responded only to individual requests for information.
“We do not provide any government organization with direct access to Facebook servers,” said Joe Sullivan, chief security officer for Facebook.
“We have never heard of PRISM,” an Apple spokesman said. “We do not provide any government agency with direct access to our servers, and any government agency requesting customer data must get a court order.”
Microsoft issued a statement Thursday saying, “We provide customer data only when we receive a legally binding order or subpoena to do so, and never on a voluntary basis.”
Google “cares deeply about the security” of its users’ data, Leslie Miller, a representative, said in a statement. “From time to time, people allege that we have created a government ‘back door’ into our systems, but Google does not have a ‘back door’ for the government to access private user data.”
Government officials and the document made clear that the NSA regarded the identities of its private partners as PRISM’s most sensitive secret, fearing that they would withdraw from the program if exposed. “98 percent of PRISM production is based on Yahoo, Google and Microsoft; we need to make sure we don’t harm these sources,” the briefing’s author wrote in his speaker’s notes.
An internal presentation of 41 briefing slides on PRISM, dated April 2013 and intended for senior analysts in the NSA’s Signals Intelligence Directorate, described the new tool as the most prolific contributor to the President’s Daily Brief, which cited PRISM data in 1,477 articles last year. According to the slides and other supporting materials, “NSA reporting increasingly relies on PRISM” as its leading source of raw material, accounting for nearly one in seven intelligence reports.
That is a remarkable figure in an agency that measures annual intake in the trillions of communications. It is all the more striking because the NSA, whose lawful mission is foreign intelligence, is reaching deep inside the machinery of U.S. companies that host hundreds of millions of U.S.-held accounts on U.S. soil.
The technology companies, which knowingly participate in PRISM operations, include most of the dominant global players of Silicon Valley, according to the document. They are listed on a roster that bears their logos in order of entry into the program: “Microsoft, Yahoo, Google, Facebook, PalTalk, AOL, Skype, YouTube, Apple.” PalTalk, although much smaller, has hosted significant traffic during the Arab Spring and in the Syrian civil war.
Government officials declined to comment for this article.
“I would just push back on the idea that the court has signed off on it, so why worry?” said Jameel Jaffer, deputy legal director of the American Civil Liberties Union (ACLU). “This is a court that meets in secret, allows only the government to appear before it, and publishes almost none of its opinions. It has never been an effective check on government.”
Internet choke points
PRISM is an heir, in one sense, to a history of intelligence alliances with as many as 100 trusted U.S. companies since the 1970s. The NSA calls these Special Source Operations, and PRISM falls under that rubric.
The Silicon Valley operation works alongside a parallel program, code-named BLARNEY, that gathers up “metadata” — address packets, device signatures and the like — as it streams past choke points along the backbone of the Internet.
The PRISM program is not a dragnet, exactly. From inside a company’s data stream the NSA is capable of pulling out anything it likes, but under current rules the agency does not try to collect it all.
Analysts who use the system from a Web portal in Fort Meade, Md., key in “selectors,” or search terms, that are designed to produce at least 51 percent confidence in a target’s “foreignness.” That is not a very stringent test. Training materials instruct new analysts to submit accidentally collected U.S. content for a quarterly report but add that “it’s nothing to worry about.”
Even when the system works as advertised, with no American singled out for targeting, the NSA routinely collects a great deal of American content. That is described as “incidental,” and it is inherent in contact chaining, one of the basic tools of the trade.
To collect data on a suspected spy or foreign terrorist means, at minimum, that everyone in the suspect’s inbox or outbox is swept in.
Firsthand experience with these systems and horror at their capabilities are what drove a career intelligence officer to provide PowerPoint slides about PRISM and supporting materials to The Washington Post to expose what he believes to be a gross intrusion on privacy.
“They quite literally can watch your ideas form as you type,” the officer said.
Washington Post researcher Julie Tate and Post staff writer Robert O’Harrow Jr. contributed to this report.