Group items tagged

GCHQ vacuums up the website browsing histories using “probes” that tap into the international fiber-optic cables that transport Internet traffic across the world. A huge volume of the Internet data GCHQ collects flows directly into a massive repository named Black Hole, which is at the core of the agency’s online spying operations, storing raw logs of intercepted material before it has been subject to analysis. Black Hole contains data collected by GCHQ as part of bulk “unselected” surveillance, meaning it is not focused on particular “selected” targets and instead includes troves of data indiscriminately swept up about ordinary people’s online activities. Between August 2007 and March 2009, GCHQ documents say that Black Hole was used to store more than 1.1 trillion “events”  a term the agency uses to refer to metadata records  with about 10 billion new entries added every day. As of March 2009, the largest slice of data Black Hole held  41 percent  was about people’s Internet browsing histories. The rest included a combination of email and instant messenger records, details about search engine queries, information about social media activity, logs related to hacking operations, and data on people’s use of tools to browse the Internet anonymously.

Throughout this period, as smartphone sales started to boom, the frequency of people’s Internet use was steadily increasing. In tandem, British spies were working frantically to bolster their spying capabilities, with plans afoot to expand the size of Black Hole and other repositories to handle an avalanche of new data. By 2010, according to the documents, GCHQ was logging 30 billion metadata records per day. By 2012, collection had increased to 50 billion per day, and work was underway to double capacity to 100 billion. The agency was developing “unprecedented” techniques to perform what it called “population-scale” data mining, monitoring all communications across entire countries in an effort to detect patterns or behaviors deemed suspicious. It was creating what it saidwould be, by 2013, “the world’s biggest” surveillance engine “to run cyber operations and to access better, more valued data for customers to make a real world difference.” HERE WAS A SIMPLE AIM at the heart of the top-secret program: Record the website browsing habits of “every visible user on the Internet.” Before long, billions of digital records about ordinary people’s online activities were being stored every day. Among them were details cataloging visits to porn, social media and news websites, search engines, chat forums, and blogs.

The mass surveillance operation — code-named KARMA POLICE — was launched by British spies about seven years ago without any public debate or scrutiny. It was just one part of a giant global Internet spying apparatus built by the United Kingdom’s electronic eavesdropping agency, Government Communications Headquarters, or GCHQ. The revelations about the scope of the British agency’s surveillance are contained in documents obtained by The Intercept from National Security Agency whistleblower Edward Snowden. Previous reports based on the leaked files have exposed how GCHQ taps into Internet cables to monitor communications on a vast scale, but many details about what happens to the data after it has been vacuumed up have remained unclear.

A huge volume of the Internet data GCHQ collects flows directly into a massive repository named Black Hole, which is at the core of the agency’s online spying operations, storing raw logs of intercepted material before it has been subject to analysis. Black Hole contains data collected by GCHQ as part of bulk “unselected” surveillance, meaning it is not focused on particular “selected” targets and instead includes troves of data indiscriminately swept up about ordinary people’s online activities. Between August 2007 and March 2009, GCHQ documents say that Black Hole was used to store more than 1.1 trillion “events” — a term the agency uses to refer to metadata records — with about 10 billion new entries added every day. As of March 2009, the largest slice of data Black Hole held — 41 percent — was about people’s Internet browsing histories. The rest included a combination of email and instant messenger records, details about search engine queries, information about social media activity, logs related to hacking operations, and data on people’s use of tools to browse the Internet anonymously.

Throughout this period, as smartphone sales started to boom, the frequency of people’s Internet use was steadily increasing. In tandem, British spies were working frantically to bolster their spying capabilities, with plans afoot to expand the size of Black Hole and other repositories to handle an avalanche of new data. By 2010, according to the documents, GCHQ was logging 30 billion metadata records per day. By 2012, collection had increased to 50 billion per day, and work was underway to double capacity to 100 billion. The agency was developing “unprecedented” techniques to perform what it called “population-scale” data mining, monitoring all communications across entire countries in an effort to detect patterns or behaviors deemed suspicious. It was creating what it said would be, by 2013, “the world’s biggest” surveillance engine “to run cyber operations and to access better, more valued data for customers to make a real world difference.”

A document from the GCHQ target analysis center (GTAC) shows the Black Hole repository’s structure.

The data is searched by GCHQ analysts in a hunt for behavior online that could be connected to terrorism or other criminal activity. But it has also served a broader and more controversial purpose — helping the agency hack into European companies’ computer networks. In the lead up to its secret mission targeting Netherlands-based Gemalto, the largest SIM card manufacturer in the world, GCHQ used MUTANT BROTH in an effort to identify the company’s employees so it could hack into their computers. The system helped the agency analyze intercepted Facebook cookies it believed were associated with Gemalto staff located at offices in France and Poland. GCHQ later successfully infiltrated Gemalto’s internal networks, stealing encryption keys produced by the company that protect the privacy of cell phone communications.

Similarly, MUTANT BROTH proved integral to GCHQ’s hack of Belgian telecommunications provider Belgacom. The agency entered IP addresses associated with Belgacom into MUTANT BROTH to uncover information about the company’s employees. Cookies associated with the IPs revealed the Google, Yahoo, and LinkedIn accounts of three Belgacom engineers, whose computers were then targeted by the agency and infected with malware. The hacking operation resulted in GCHQ gaining deep access into the most sensitive parts of Belgacom’s internal systems, granting British spies the ability to intercept communications passing through the company’s networks.

In March, a U.K. parliamentary committee published the findings of an 18-month review of GCHQ’s operations and called for an overhaul of the laws that regulate the spying. The committee raised concerns about the agency gathering what it described as “bulk personal datasets” being held about “a wide range of people.” However, it censored the section of the report describing what these “datasets” contained, despite acknowledging that they “may be highly intrusive.” The Snowden documents shine light on some of the core GCHQ bulk data-gathering programs that the committee was likely referring to — pulling back the veil of secrecy that has shielded some of the agency’s most controversial surveillance operations from public scrutiny. KARMA POLICE and MUTANT BROTH are among the key bulk collection systems. But they do not operate in isolation — and the scope of GCHQ’s spying extends far beyond them.

The agency operates a bewildering array of other eavesdropping systems, each serving its own specific purpose and designated a unique code name, such as: SOCIAL ANTHROPOID, which is used to analyze metadata on emails, instant messenger chats, social media connections and conversations, plus “telephony” metadata about phone calls, cell phone locations, text and multimedia messages; MEMORY HOLE, which logs queries entered into search engines and associates each search with an IP address; MARBLED GECKO, which sifts through details about searches people have entered into Google Maps and Google Earth; and INFINITE MONKEYS, which analyzes data about the usage of online bulletin boards and forums. GCHQ has other programs that it uses to analyze the content of intercepted communications, such as the full written body of emails and the audio of phone calls. One of the most important content collection capabilities is TEMPORA, which mines vast amounts of emails, instant messages, voice calls and other communications and makes them accessible through a Google-style search tool named XKEYSCORE.

As of September 2012, TEMPORA was collecting “more than 40 billion pieces of content a day” and it was being used to spy on people across Europe, the Middle East, and North Africa, according to a top-secret memo outlining the scope of the program. The existence of TEMPORA was first revealed by The Guardian in June 2013. To analyze all of the communications it intercepts and to build a profile of the individuals it is monitoring, GCHQ uses a variety of different tools that can pull together all of the relevant information and make it accessible through a single interface. SAMUEL PEPYS is one such tool, built by the British spies to analyze both the content and metadata of emails, browsing sessions, and instant messages as they are being intercepted in real time. One screenshot of SAMUEL PEPYS in action shows the agency using it to monitor an individual in Sweden who visited a page about GCHQ on the U.S.-based anti-secrecy website Cryptome.

Partly due to the U.K.’s geographic location — situated between the United States and the western edge of continental Europe — a large amount of the world’s Internet traffic passes through its territory across international data cables. In 2010, GCHQ noted that what amounted to “25 percent of all Internet traffic” was transiting the U.K. through some 1,600 different cables. The agency said that it could “survey the majority of the 1,600” and “select the most valuable to switch into our processing systems.”

According to Joss Wright, a research fellow at the University of Oxford’s Internet Institute, tapping into the cables allows GCHQ to monitor a large portion of foreign communications. But the cables also transport masses of wholly domestic British emails and online chats, because when anyone in the U.K. sends an email or visits a website, their computer will routinely send and receive data from servers that are located overseas. “I could send a message from my computer here [in England] to my wife’s computer in the next room and on its way it could go through the U.S., France, and other countries,” Wright says. “That’s just the way the Internet is designed.” In other words, Wright adds, that means “a lot” of British data and communications transit across international cables daily, and are liable to be swept into GCHQ’s databases.

A map from a classified GCHQ presentation about intercepting communications from undersea cables. GCHQ is authorized to conduct dragnet surveillance of the international data cables through so-called external warrants that are signed off by a government minister. The external warrants permit the agency to monitor communications in foreign countries as well as British citizens’ international calls and emails — for example, a call from Islamabad to London. They prohibit GCHQ from reading or listening to the content of “internal” U.K. to U.K. emails and phone calls, which are supposed to be filtered out from GCHQ’s systems if they are inadvertently intercepted unless additional authorization is granted to scrutinize them. However, the same rules do not apply to metadata. A little-known loophole in the law allows GCHQ to use external warrants to collect and analyze bulk metadata about the emails, phone calls, and Internet browsing activities of British people, citizens of closely allied countries, and others, regardless of whether the data is derived from domestic U.K. to U.K. communications and browsing sessions or otherwise. In March, the existence of this loophole was quietly acknowledged by the U.K. parliamentary committee’s surveillance review, which stated in a section of its report that “special protection and additional safeguards” did not apply to metadata swept up using external warrants and that domestic British metadata could therefore be lawfully “returned as a result of searches” conducted by GCHQ.

Perhaps unsurprisingly, GCHQ appears to have readily exploited this obscure legal technicality. Secret policy guidance papers issued to the agency’s analysts instruct them that they can sift through huge troves of indiscriminately collected metadata records to spy on anyone regardless of their nationality. The guidance makes clear that there is no exemption or extra privacy protection for British people or citizens from countries that are members of the Five Eyes, a surveillance alliance that the U.K. is part of alongside the U.S., Canada, Australia, and New Zealand. “If you are searching a purely Events only database such as MUTANT BROTH, the issue of location does not occur,” states one internal GCHQ policy document, which is marked with a “last modified” date of July 2012. The document adds that analysts are free to search the databases for British metadata “without further authorization” by inputing a U.K. “selector,” meaning a unique identifier such as a person’s email or IP address, username, or phone number. Authorization is “not needed for individuals in the U.K.,” another GCHQ document explains, because metadata has been judged “less intrusive than communications content.” All the spies are required to do to mine the metadata troves is write a short “justification” or “reason” for each search they conduct and then click a button on their computer screen.

Intelligence GCHQ collects on British persons of interest is shared with domestic security agency MI5, which usually takes the lead on spying operations within the U.K. MI5 conducts its own extensive domestic surveillance as part of a program called DIGINT (digital intelligence).

GCHQ’s documents suggest that it typically retains metadata for periods of between 30 days to six months. It stores the content of communications for a shorter period of time, varying between three to 30 days. The retention periods can be extended if deemed necessary for “cyber defense.” One secret policy paper dated from January 2010 lists the wide range of information the agency classes as metadata — including location data that could be used to track your movements, your email, instant messenger, and social networking “buddy lists,” logs showing who you have communicated with by phone or email, the passwords you use to access “communications services” (such as an email account), and information about websites you have viewed.

Records showing the full website addresses you have visited — for instance, www.gchq.gov.uk/what_we_do — are treated as content. But the first part of an address you have visited — for instance, www.gchq.gov.uk — is treated as metadata. In isolation, a single metadata record of a phone call, email, or website visit may not reveal much about a person’s private life, according to Ethan Zuckerman, director of Massachusetts Institute of Technology’s Center for Civic Media. But if accumulated and analyzed over a period of weeks or months, these details would be “extremely personal,” he told The Intercept, because they could reveal a person’s movements, habits, religious beliefs, political views, relationships, and even sexual preferences. For Zuckerman, who has studied the social and political ramifications of surveillance, the most concerning aspect of large-scale government data collection is that it can be “corrosive towards democracy” — leading to a chilling effect on freedom of expression and communication. “Once we know there’s a reasonable chance that we are being watched in one fashion or another it’s hard for that not to have a ‘panopticon effect,’” he said, “where we think and behave differently based on the assumption that people may be watching and paying attention to what we are doing.”

When compared to surveillance rules in place in the U.S., GCHQ notes in one document that the U.K. has “a light oversight regime.” The more lax British spying regulations are reflected in secret internal rules that highlight greater restrictions on how NSA databases can be accessed. The NSA’s troves can be searched for data on British citizens, one document states, but they cannot be mined for information about Americans or other citizens from countries in the Five Eyes alliance. No such constraints are placed on GCHQ’s own databases, which can be sifted for records on the phone calls, emails, and Internet usage of Brits, Americans, and citizens from any other country. The scope of GCHQ’s surveillance powers explain in part why Snowden told The Guardian in June 2013 that U.K. surveillance is “worse than the U.S.” In an interview with Der Spiegel in July 2013, Snowden added that British Internet cables were “radioactive” and joked: “Even the Queen’s selfies to the pool boy get logged.”

In recent years, the biggest barrier to GCHQ’s mass collection of data does not appear to have come in the form of legal or policy restrictions. Rather, it is the increased use of encryption technology that protects the privacy of communications that has posed the biggest potential hindrance to the agency’s activities. “The spread of encryption … threatens our ability to do effective target discovery/development,” says a top-secret report co-authored by an official from the British agency and an NSA employee in 2011. “Pertinent metadata events will be locked within the encrypted channels and difficult, if not impossible, to prise out,” the report says, adding that the agencies were working on a plan that would “(hopefully) allow our Internet Exploitation strategy to prevail.”

We're also concerned that Verizon's failure to permit its users to opt out of X-UIDH may be a violation of the federal law that requires phone companies to maintain the confidentiality of their customers' data. Only two months ago, the wireline sector of Verizon's business was hit with a $7.4 million fine by the Federal Communications Commission after it was caught using its "customers' personal information for thousands of marketing campaigns without even giving them the choice to opt out." With this header, it looks like Verizon lets its customers opt out of the marketing side of the program, but not from the disclosure of their browsing habits.

He agreed to a video chat that turned into a three-hour discussion sprawling from the ethics of surveillance to the downsides of home improvements and the difficulty of securing your laptop.

In recent years, two developments have helped make hacking for the government a lot more attractive than hacking for yourself. First, the Department of Justice has cracked down on freelance hacking, whether it be altruistic or malignant. If the DOJ doesn’t like the way you hack, you are going to jail. Meanwhile, hackers have been warmly invited to deploy their transgressive impulses in service to the homeland, because the NSA and other federal agencies have turned themselves into licensed hives of breaking into other people’s computers. For many, it’s a techno sandbox of irresistible delights, according to Gabriella Coleman, a professor at McGill University who studies hackers. “The NSA is a very exciting place for hackers because you have unlimited resources, you have some of the best talent in the world, whether it’s cryptographers or mathematicians or hackers,” she said. “It is just too intellectually exciting not to go there.”

The Lamb’s memos on cool ways to hunt sysadmins triggered a strong reaction when I wrote about them in 2014 with my colleague Ryan Gallagher. The memos explained how the NSA tracks down the email and Facebook accounts of systems administrators who oversee computer networks. After plundering their accounts, the NSA can impersonate the admins to get into their computer networks and pilfer the data flowing through them. As the Lamb wrote, “sys admins generally are not my end target. My end target is the extremist/terrorist or government official that happens to be using the network … who better to target than the person that already has the ‘keys to the kingdom’?” Another of his NSA memos, “Network Shaping 101,” used Yemen as a theoretical case study for secretly redirecting the entirety of a country’s internet traffic to NSA servers.

“If I turn the tables on you,” I asked the Lamb, “and say, OK, you’re a target for all kinds of people for all kinds of reasons. How do you feel about being a target and that kind of justification being used to justify getting all of your credentials and the keys to your kingdom?” The Lamb smiled. “There is no real safe, sacred ground on the internet,” he replied. “Whatever you do on the internet is an attack surface of some sort and is just something that you live with. Any time that I do something on the internet, yeah, that is on the back of my mind. Anyone from a script kiddie to some random hacker to some other foreign intelligence service, each with their different capabilities — what could they be doing to me?”

“You know, the situation is what it is,” he said. “There are protocols that were designed years ago before anybody had any care about security, because when they were developed, nobody was foreseeing that they would be taken advantage of. … A lot of people on the internet seem to approach the problem [with the attitude of] ‘I’m just going to walk naked outside of my house and hope that nobody looks at me.’ From a security perspective, is that a good way to go about thinking? No, horrible … There are good ways to be more secure on the internet. But do most people use Tor? No. Do most people use Signal? No. Do most people use insecure things that most people can hack? Yes. Is that a bash against the intelligence community that people use stuff that’s easily exploitable? That’s a hard argument for me to make.”

I mentioned that lots of people, including Snowden, are now working on the problem of how to make the internet more secure, yet he seemed to do the opposite at the NSA by trying to find ways to track and identify people who use Tor and other anonymizers. Would he consider working on the other side of things? He wouldn’t rule it out, he said, but dismally suggested the game was over as far as having a liberating and safe internet, because our laptops and smartphones will betray us no matter what we do with them. “There’s the old adage that the only secure computer is one that is turned off, buried in a box ten feet underground, and never turned on,” he said. “From a user perspective, someone trying to find holes by day and then just live on the internet by night, there’s the expectation [that] if somebody wants to have access to your computer bad enough, they’re going to get it. Whether that’s an intelligence agency or a cybercrimes syndicate, whoever that is, it’s probably going to happen.”

There are precautions one can take, and I did that with the Lamb. When we had our video chat, I used a computer that had been wiped clean of everything except its operating system and essential applications. Afterward, it was wiped clean again. My concern was that the Lamb might use the session to obtain data from or about the computer I was using; there are a lot of things he might have tried, if he was in a scheming mood. At the end of our three hours together, I mentioned to him that I had taken these precautions—and he approved. “That’s fair,” he said. “I’m glad you have that appreciation. … From a perspective of a journalist who has access to classified information, it would be remiss to think you’re not a target of foreign intelligence services.” He was telling me the U.S. government should be the least of my worries. He was trying to help me. Documents published with this article: Tracking Targets Through Proxies & Anonymizers Network Shaping 101 Shaping Diagram I Hunt Sys Admins (first published in 2014)

Starting in 2010, EFF members have pushed tech companies to follow crypto best practices. We applauded when Facebook and Twitter implemented HTTPS by default, and when Wikipedia and several other popular sites later followed suit. Google has also put pressure on the tech community by using HTTPS as a signal in search ranking algorithms and, starting this year, showing security warnings in Chrome when users load HTTP sites that request passwords or credit card numbers. EFF’s Encrypt the Web Report also played a big role in tracking and encouraging specific practices. Recently other organizations have followed suit with more sophisticated tracking projects. For example, Secure the News and Pulse track HTTPS progress among news media sites and U.S. government sites, respectively.

But securing large, popular websites is only one part of a much bigger battle. Encrypting the entire web requires HTTPS implementation to be accessible to independent, smaller websites. Let’s Encrypt and Certbot have changed the game here, making what was once an expensive, technically demanding process into an easy and affordable task for webmasters across a range of resource and skill levels. Let’s Encrypt is a Certificate Authority (CA) run by the Internet Security Research Group (ISRG) and founded by EFF, Mozilla, and the University of Michigan, with Cisco and Akamai as founding sponsors. As a CA, Let’s Encrypt issues and maintains digital certificates that help web users and their browsers know they’re actually talking to the site they intended to. CAs are crucial to secure, HTTPS-encrypted communication, as these certificates verify the association between an HTTPS site and a cryptographic public key. Through EFF’s Certbot tool, webmasters can get a free certificate from Let’s Encrypt and automatically configure their server to use it. Since we announced that Let’s Encrypt was the web’s largest certificate authority last October, it has exploded from 12 million certs to over 28 million. Most of Let’s Encrypt’s growth has come from giving previously unencrypted sites their first-ever certificates. A large share of these leaps in HTTPS adoption are also thanks to major hosting companies and platforms--like WordPress.com, Squarespace, and dozens of others--integrating Let’s Encrypt and providing HTTPS to their users and customers.

Unfortunately, you can only use HTTPS on websites that support it--and about half of all web traffic is still with sites that don’t. However, when sites partially support HTTPS, users can step in with the HTTPS Everywhere browser extension. A collaboration between EFF and the Tor Project, HTTPS Everywhere makes your browser use HTTPS wherever possible. Some websites offer inconsistent support for HTTPS, use unencrypted HTTP as a default, or link from secure HTTPS pages to unencrypted HTTP pages. HTTPS Everywhere fixes these problems by rewriting requests to these sites to HTTPS, automatically activating encryption and HTTPS protection that might otherwise slip through the cracks.

Our goal is a universally encrypted web that makes a tool like HTTPS Everywhere redundant. Until then, we have more work to do. Protect your own browsing and websites with HTTPS Everywhere and Certbot, and spread the word to your friends, family, and colleagues to do the same. Together, we can encrypt the entire web.