Apple, Microsoft, Adobe, Oracle programs vulnerable to cyberattacks, analysis finds
For years, fixing the software flaws that left Adobe Systems Inc.'s customers prey to hackers simply wasn't a top company priority.
At Adobe, whose Acrobat and Reader programs for creating and viewing PDF documents sit on most of the planet's personal computers, "security never made the list," says Brad Arkin, senior director of security for products and services at the San Jose, Calif.-based company.
Then, in 2009, what became known as the JBIG2 flaw created an existential crisis for the company. For the first time, hackers were found to be using a crack in the armor of Acrobat and Reader to infiltrate major corporations and Adobe had no available fix.
"That was the big wake-up call," Arkin says. "We needed to make some big changes to protect our users."
The company realized it had to defend itself against a new, sophisticated type of hacker-spy targeting the software's corporate users and their secrets, Arkin says.
In one instance, attackers used the JBIG2 defect to get access to the computer of a Coca-Cola Co. executive in China through an infected Adobe PDF they e-mailed to her, according to an internal Coca-Cola document obtained by Bloomberg. The beverage maker was involved in what would have been the largest foreign takeover of a Chinese company at the time.
After finding there were too many imperfections to fix, Arkin says he instead erected a virtual wall around the programs, and focused on keeping that defense intact.
It might not be enough.
"Imagine a castle wall as long as the Great Wall of China," says Kyle Randolph, a former senior manager of product security at Adobe, who worked there from 2008 until this year. "All you need is one hole and the whole thing is compromised."
Flaws in the ubiquitous software on PCs, tablets and smartphones have empowered cyber intruders and plagued businesses, governments and political dissidents with sabotage, theft and physical attacks, a year-long series by Bloomberg News shows. In part, it is the legacy of companies that eschewed vigilance, putting profit before safety.
That it took a series of attacks on corporate customers to shock the U.S. software company into making security a priority helps illustrate why today -- two decades after the Internet age took off -- computer users are at risk whenever and wherever they're online.
Products used on virtually all computers, from Adobe, Apple Inc., Microsoft Corp. and Oracle Corp., consistently dominate industry rankings of programs most vulnerable to attack. The resulting Swiss cheese of imperfections has made every citizen a potential crack in the security walls meant to protect their governments, employers and anyone with whom they do business.
Across the industry, software makers say they are taking security seriously and making improvements to address the increasingly sophisticated hacker threat. For instance, Microsoft and Adobe have made it easier for users to get updates that patch defects, and Google Inc. fends off attacks by encrypting traffic on its Gmail service.
Adobe's Arkin says the company's strategy makes the software easier to defend by requiring it to safeguard about 8,000 lines of code that hackers could use to breach the protective wall, instead of tens of millions of lines in the underlying programs. While the programs won't be perfect forever, Adobe is working to keep ahead of the hackers by making their jobs harder and more expensive, Arkin says. "Our end goal is to protect users of our software," he says.
The flaws have nevertheless flourished in the absence of industry standards or product liability.
Attempts to force the architects of the Internet to improve the safety of users have so far failed, in part because the U.S. Chamber of Commerce has pushed back on behalf of its business members. It helped defeat a cyber security bill backed by the White House this year that included regulation of the small fraction of corporate computer systems that, if hacked, could cause mass casualties or economic damage. Chamber lobbyists cast the bill as unnecessary and overreaching, and said government meddling would only make the problem worse.
In response to questions about its opposition to the bill, the Chamber provided a letter it sent last month to the U.S. Senate, favoring a "workable" bill focused on information sharing, and voicing "serious concerns" about government interference with the private sector.
U.S. Defense Secretary Leon Panetta invoked Pearl Harbor in an Oct. 11 speech warning that extremist groups or a rival nation might use cyber tools to attack.
In America and the U.K., about 1-in-3 computer users had contact with malicious software, just between July and September this year, according to data Moscow-based anti-virus software maker Kaspersky Lab collected from its customers.
The implications of lagging security go beyond PCs to critical infrastructure and industry, such as power grids and railroads, and to increasingly networked lives, including phone systems and videoconferencing that run over the Internet.
"Sooner or later, the people who are exploiting these security flaws will go from stealing information to breaking systems -- because they can -- and then it's going to be obvious to everybody how bad things are," says Stewart Baker, former general counsel for the National Security Agency, the U.S. spy agency, which monitors foreign communications.
Increasing the security of millions of lines of code underlying some of the world's most popular software would be time-consuming and expensive. Behind closed doors, software makers consistently argue that while consumers may appreciate more security, there is little evidence they'd sacrifice functionality, time-to-market or cost to get it, according to three policy makers who regularly attend meetings with software company chief executive officers. They asked not to be named because the meetings were confidential.
In a series of stories that showed the global abuses and costs of cyber weapons and espionage, Bloomberg News uncovered a diverse array of attackers and targets: A hacker group linked by U.S. intelligence to the Chinese military, according to a U.S diplomatic cable released by Wikileaks, stole U.S. corporate secrets and pilfered bureaucrats' e-mails in Brussels, while commercial spyware made in Europe hit Persian Gulf activists; and Syrians fought a cyber war via online chats and webmail with rudimentary tools and deadly results.
As different as the examples are from each other, a single thread runs through them all: flawed software or network design enabled the hacks.
A commercial spy program sold to governments -- and discovered by Bloomberg News to have been used in the surveillance of Bahraini democracy activists -- had advertised that it was able to gain control of computers through an Apple iTunes flaw that was publicly known for three years before Apple patched it in November 2011. The manufacturer of the FinFisher software, U.K-based Gamma Group, pins the blame on Apple.
"They could have protected users," says Martin J. Muench, the managing director of Gamma's German unit. "Security wasn't their highest priority." 'Good Impact'
Apple spokesman Bill Evans said the iTunes security problem has been fixed and declined to comment on Muench's statement.
Microsoft has taken steps to keep customers safer, says Matt Thomlinson, general manager of product security at the Redmond, Washington-based company. It ramped up the effort in 2002 with the establishment of its Trustworthy Computing initiative, which deals with security and privacy issues. Since 2004, Microsoft has made automated updates a standard setting for users of its Windows operating system.
"We've had some really good impact on driving down attacks through vulnerabilities, but that doesn't mean the attackers are going away," he says.
"I don't think you're ever going to get to the point where there are zero vulnerabilities," says Thomlinson, who has been with Microsoft since 1994. "We can remove vulnerabilities from the code, and that's what we've attempted to do."
As companies such as Microsoft harden their defenses, hackers find new ways to get in, improving their skills or going after other products, says Steven M. Christey, principal information security engineer at MITRE Corp., a non-profit group based in Massachusetts and Virginia that tracks vulnerabilities.
"Software security is a rapidly moving target," he says. "The result is a perpetual arms race between the software vendor and the attacker." In the continuous battle, an informal army of hackers, security firms, academics and other bug hunters seeks out imperfections in programs that an attacker can use to compromise a computer. Some privately alert software makers to flaws they find.
Others devise methods for using those defects -- known as exploits -- and then sell or publicize them as hacking tools.
Despite new priorities for security, some software makers give hackers head starts by failing to fix the problems quickly.
This year, Oracle didn't patch a bug in Java, its computing platform for a range of games, trading and other programs, until almost five months after researchers alerted it about the vulnerability.
On April 2, Polish firm Security Explorations told Oracle it had discovered a defect in the product. Months went by with no solutions from Oracle, the world's largest supplier of database software.
That began to change on Aug. 26, when an alert went up on a website that tracks software vulnerability. The site, run by the security firm FireEye Inc., said that a new Java flaw had been spotted in targeted attacks.
The next day, computer code for using that flaw to invade machines running Java was added to a hacking tool kit available for free online, called Metasploit. Virtually any hacker could now download the exploit and use it.
HD Moore, chief security officer at Rapid7, the Boston- based company that runs Metasploit, says publishing exploits forces companies to be accountable. "It's hitting the vendor over the head with a hammer," he says. "When everybody has access to it, they need to take it seriously."
On Aug. 28, Milpitas, Calif.-based FireEye found evidence of the "first indication of a large scale attack" with the new exploit. AlienVault, a security firm, linked the original attacks to China-based hackers.
Then, two days later, Oracle issued patches to protect customers.
Hackers have piled on, making Java the most targeted application during the third quarter of 2012, accounting for 56 percent of exploits blocked by Kaspersky Lab anti-virus software. Adobe took second place, with 28 percent.
"This year, yes, Oracle has been the worst," says Jaime Blasco, head of the security lab at AlienVault.
Oracle spokeswoman Deborah Hellinger declined to comment on the Java flaw identified by Polish researchers or its system for patching vulnerabilities.
Arkin said the Kaspersky ranking catches many attacks aimed at problems in older versions that have been fixed but which haven't been updated by users. "That's the milk carton with last year's date on it," he says. "We say, 'We hear you, we've listened and here's the solution.'"
While hacking's dangers can't be eliminated, they can be reduced, as the evolution of the auto industry has shown.
In 1965, consumer rights advocate Ralph Nader published "Unsafe at Any Speed," detailing how U.S. automakers, fearing higher costs, resisted measures such as seat belts and ignored crash-test findings. The book spawned congressional hearings and modern safety requirements -- and since then the U.S. per-capita death rate from auto accidents has been cut by more than half. Cars still crash, but new ones come with air bags.
Almost a half century later, Nader says the government and online service providers should do more to safeguard consumers from the new threat. "It does seem they can't keep up with the genius nature of the hackers," says Nader. "Basically it's the cost of doing business."
For the Internet, the current safety crossroads has roots in the fine print that comes with most software licenses, which largely absolve makers of the products underpinning the Internet of liability.
Legal liability can wreck an industry or move it offshore, Baker, the former NSA general counsel, says.
"Software security is uniquely resistant to the kind of law making and standard setting that courts do," he says.
Today the onus is on consumers to protect themselves so they don't become the conduit that lets attackers into their businesses, utilities or governments. To help reduce those risks, there are several steps technology companies can take to add virtual seat belts and air bags to their products and give customers safer choices.
In one simple solution already adopted by some suppliers, software makers can automate the patches for flaws, rather than depending on users to figure out how to do it themselves, says Stefano Ortolani, a security researcher for Kaspersky Lab Italy.
Even with the Java patch issued in August, for instance, there was no automated update process for all users, so many had to either deploy the fix themselves or disable Java to be protected. Today about 65 percent of Java users lack the latest version, which includes patches of security holes, according to Rapid7.
Service providers and software makers can change the ways people use computer programs by keeping applications off devices altogether, and running them in a new network of remote computers, known as the cloud. With word processing or accounting programs in the cloud, rather than individual machines, software makers can ensure security fixes are applied for all users and are done more quickly, says Rob Rachwald, director of security strategy at Imperva Inc., a Redwood Shores, Calif.-based web security company.
"You're using it on the Web, not your desktop," he says.
Software makers could provide another fix that would deprive hackers of the predictable roadmaps they use to navigate peoples' computers by purposely jumbling where programs and data are stored in a machine's memory. That defensive move doesn't stop hackers from getting in, but can keep them from taking control, says Andy Chou, chief technology officer of San Francisco-based Coverity Inc., which finds and fixes software defects for clients.
Consumers themselves have options to mitigate risk, some borrowed from the world of hacking, from encrypting their e- mails to using special Web browsers that disguise their identities.
Using sites whose addresses start with HTTPS -- an encrypted version of HTTP -- adds a layer of protection. Most webmail providers, such as Google's Gmail, have made HTTPS standard, while Yahoo! Inc. until recently had held out. The San Francisco-based Electronic Frontier Foundation has said Yahoo puts users at risk, especially in repressive countries.
"They deserve to be shamed loudly and at length," EFF's International Freedom of Expression Coordinator Eva Galperin said in a Nov. 28 e-mail.
In response to questions about EFF's position, Yahoo said that it has begun offering the technology behind HTTPS and will continue to do so over the next few months, and that it is developing and testing more secure platforms.
"Yahoo! is committed to protecting the security and privacy of our users," the statement said. The EFF's Galperin said, "I am glad they are finally getting started." To achieve the cyber equivalent of crash-testing, industry standards for vetting programs would tip the balance toward safety, says Sean Coyne, a Washington-based cyber-security researcher who has worked for government and corporate clients.
"Nobody is holding the manufacturers' feet to the fire," Coyne says.
Minimum security standards might lead software makers to beef up the now paltry ranks of programmers dedicated to making sure products are safe. It's not unusual for software makers to have one security person per every 500 to 1,000 developers, says Coverity's Chou.
"Those 1,000 developers are writing code every day -- how can that one poor guy keep up?" he says.
As corporate customers fend for themselves, they also can do more to keep safe. Many of the hacks uncovered this year by Bloomberg News showed basic -- and fixable -- flaws in internal security.
In the 2009 Coca-Cola hack, an internal investigation found that it was the work of state-backed hackers, and that the company neglected to take several protective steps. Details in a company document detailing the attack indicate the intruders were part of a prolific China-based hacking group, according to researchers interviewed by Bloomberg.
Coca-Cola needed to patch Adobe Reader vulnerabilities to address immediate holes and, for the longer term, establish a comprehensive program for managing fixes, according to the report.
When inadequate records prevented them from reconstructing the hack, investigators recommended the company implement systems to perform centralized logging of what occurs on the network. Coca-Cola spokesman Kent Landers declined to comment when asked this week about any follow-up to the incident.
The story behind the Coca-Cola hack is also the tale of Adobe's fight against the JBIG2 bug. Adobe's challenge, as told by former and current employees, provides a window into the tension between profit and safety that has shaped the Internet.
Tom Ferris, an Adobe researcher from 2006 to 2010, worked finding security failures and triaging all the security vulnerabilities streaming in. He recalls that when he started, they'd find 10 to 15 flaws a week that needed a fix. "I told my managers this was a big problem," he says.
They ignored him, until the JBIG2 bug hit, he says. The flaw causes the Adobe Reader and Acrobat applications to crash, allowing an attacker to take control of the system.
"Everything changed," Ferris says.
The ordeal began on Feb. 19, 2009, when researchers at Symantec Corp., the world's biggest security-software maker, said they'd discovered the bug, and alerted Adobe. That day, Adobe issued a security warning, saying a fix wouldn't be available until the following month.
In China, and possibly elsewhere, hackers pounced. On March 3 and 4, 2009, U.S. defense contractor Lockheed Martin Corp.'s incident response team intercepted intrusion attempts that used the vulnerability in e-mails purporting to be about an aeronautics conference and a meeting on U.S. missile defense, according to a paper the company researchers wrote.
It took until March 10 for Adobe to come out with an updated program that users could download to mitigate the flaw.
Three days later, hackers hit an executive in China for Coca-Cola, which was awaiting approval for its $2.4 billion acquisition of China Huiyuan Juice Group. It was one of several hacks using different means that compromised company executives at the time.
On March 18, 2009, the Chinese Ministry of Commerce rejected Coca-Cola's acquisition citing antitrust grounds.
Inside Adobe, the security push had just begun. Yet fixing every line of the programs wasn't in the cards because the company wasn't willing to take on the expense, says Ferris and his former colleague, Randolph.
"It's just not economical," Randolph says. "They're still making money and they're doing what's best for their shareholders." Arkin says the company heavily automated its process for releasing a patch, reducing the time lag from weeks to an average of five days. They turned off risky features, leaving it to users to decide if they wanted them.
The best bet, Arkin says he decided, was to find a way to quarantine programs. In the design, called sandboxing, a hacker might manage to take over Adobe's Reader, but would need to escape the virtual sandbox to do further damage on a user's computer.
In November 2010, Adobe released Reader X, promoting the sandbox as a way to better protect customers. "To date, we're not aware of any customer anywhere who is running Adobe Reader X who has ever been attacked in the real world," Arkin says.
That's the kind of challenge hackers love.