IRS breach highlights weakness of ‘knowledge-based’ security

The compromise of 100,000 taxpayer accounts through the Get Transcript application on the IRS website were not random hacks, but the exploits of an already-publicized vulnerability -- known to security experts since at least March.

And, in order to gain access, hackers already had a good deal of information on the affected taxpayers.

Hackers already had the keys

"We're confident that these are not amateurs," IRS Commissioner John Koskinen told the Associated Press, blaming the breach on organized crime. "These are extremely sophisticated criminals with access to a tremendous amount of data."

Taxpayers' prior-year tax filings, containing personal information from marital status to Social Security numbers to adjusted gross income, were exposed to hackers.

But hackers didn't worm through a digital backdoor to fraudulently gain access to tax returns; they simply unlocked the main entrance using information they already had or could readily guess.

"In this sophisticated effort, third parties succeeded in clearing a multi-step authentication process that required prior personal knowledge about the taxpayer, including Social Security information, date of birth, tax filing status and street address before accessing IRS systems," said the IRS statement on the breach. "The multi-layer process also requires an additional step, where applicants must correctly answer several personal identity verification questions that typically are only known by the taxpayer."

Where did hackers get that personal information?

The weakness of KBA

Security expert Brian Krebs had warned of the vulnerability of Get Transcript in mid-March by pointing out the obvious: It's not that hard for people to find the "personal identity verification" information needed to get past the IRS's authentication process.

"The IRS's process for verifying people requesting transcripts is vulnerable to exploitation by fraudsters because it relies on static identifiers and so-called "knowledge-based authentication" (KBA) -- i.e., challenge questions that can be easily defeated with information widely available for sale in the cybercrime underground and/or with a small amount of searching online," Krebs wrote.

Krebs shared the story of a man who found his SSN had been used to file a return he knew nothing about, and who then learned someone had registered using his SSN and an unknown email address on the IRS site.

The man said he believed the scammer got the information to file a false return through the Get Transcript app.

Krebs noted that "the KBA questions — which involve multiple choice, 'out of wallet' questions such as previous address, loan amounts and dates — can be successfully enumerated with random guessing." But he and International Computer Science Institute researcher Nicholas Weaver found that services such as Zillow and Spokeo supplied basically all of the KBA answers.

Between the power of search engines and the growing numbers of personal information-revealing data breaches -- from Anthem health insurance to the Postal Service, not to mention rampant state-level tax fraud and lax security at the third-party tax preparation firm on which tens of millions of Americans rely -- KBA just doesn't cut it anymore.

"We live in a world where the Internet has become a database of 'you' and where one data breach can easily feed another," said Ken Westin, security analysts with Tripwire. "The information that was used to bypass the security screen, including Social Security numbers, dates of birth and street addresses, are all components of data that have recently been compromised in health insurance data breaches."

"Tax filing status can be identified pretty easily if you know whether the person is married or not," Westin continued. "Unfortunately, the high number of large-scale data breaches has essentially transformed our personal information into public information; and this data should not be used as security or authentication checks."

"The underlying weakness in the IRS and other government website portals is they rely on knowledge-based authentication," said Brad Taylor, president and CEO of managed security service provider Proficio in a statement issued after the IRS disclosed the attack. "IRS needs to add more context to their challenge questions and monitor attempted access for suspicious behavior like multiple sign-ups from the same IP address."

What happens next

The IRS noted that only half of the roughly 200,000 attempts to access taxpayer information from questionable email domains were successful, and that the fraud is a small figure compared to the number of legitimate downloads.

"During this filing season, taxpayers successfully and safely downloaded a total of approximately 23 million transcripts," the agency noted. "The IRS emphasizes this incident involves one application involving transcripts — it does not involve other IRS systems, such as our core taxpayer accounts or other applications, such as Where's My Refund."

The IRS estimated it paid out nearly $6 billion in fraudulent 2013 refunds to identity thieves, while in this latest breach, Koskinen estimated that less than $50 million was actually paid out to scammers.

But with the wealth of information revealed on a tax return, it's possible identity thieves will save the info to use on future fraudulent tax returns, or pursue different forms of identity theft altogether.

Jonathan Sander, strategy and research officer at cybersecurity and Internet threat detection/prevention provider STEALTHbits, said it's likely scammers will do a "value-add" on the Get Transcript info, combining it with other stolen information to resell.

Combining it with, say stolen banking data, he said, could yield a premium on the growing dark web data markets.

The New York Times noted that the breach might spur President Barack Obama to push for an increase in IRS funding, though experts at the Government Accountability Office have said the IRS's IT problems stem less from a lack of funding than the lack of a digital strategy.

An IRS criminal investigation into the latest breach, along with an agency inspector general investigation, is underway.

In the meantime, the IRS will be sending letters to the affected taxpayers, both the 100,000 whose information was exposed and the 100,000 unsuccessfully targeted, and it will pay for credit monitoring for the former group.

The Get Transcript web app has been shut down.

Mark Rockwell contributed to this story.