Google Docs phishing attack underscores OAuth security risks

In this case, the culprit chose to name the app “Google Docs” in an effort to trick users.

The search company has shut down the attack by removing the app. It’s also barred other developers from using “Google” in naming their third-party apps.

However, Austin found he could still reproduce Wednesday’s phishing scheme. He did so, by using the search company’s developer platform to create his own third-party app, and also called it “Google Docs.”

The only difference is that Austin used a Cyrillic character, used in Russia, for the letter “o” in his app’s name.

“The Cyrillic letter o looks exactly like the other letter o,” Austin said. He then replicated the rest of the Wednesday’s attack, creating a fake email that uses the same design interface.  

Austin has submitted the security issue to Google, and now its developer platform no longer accepts apps under that name. However, he and other security experts predict that bad actors are also working on replicating Wednesday’s attack.

“There’s no question that this will be repeated again,” said Ayse Kaya, a director at Cisco Cloudlock Cyberlabs, a security provider. “It will probably happen much more often.”

More traditional phishing email schemes can strike by tricking users into giving up their login credentials. However, Wednesday’s attack takes a different approach and abuses what’s known as the OAuth protocol, a convenient way for internet accounts to link with third-party applications.

Through OAuth, users don’t have to hand over any password information. They instead grant permission so that one third-party app can connect to their internet account, at say, Google, Facebook or Twitter.

[ Related: What is OAuth? How the open authorization framework works ]

But like any technology, OAuth can be exploited. Back in 2011, one developer even warned that the protocol could be used in a phishing attack with apps that impersonate Google services.

Nevertheless, OAuth has become a popular standard used across IT. CloudLock has found that over 276,000 apps use the protocol through services like Google, Facebook and Microsoft Office 365.

What aided Wednesday’s phishing scheme was that Google’s own services didn’t do enough to point out it came from a suspicious developer, said Aaron Parecki, an IT consultant who helps businesses implement OAuth.

For instance, the dummy Google Docs app was registered to a developer at eugene.pupov@gmail.com — a red flag that the product wasn’t real.

However, the dummy app still managed to fool users because Google’s own account permission page never plainly listed the developer’s information, unless the user clicks the page to find out, Parecki said.

“I was surprised Google didn’t show much identifying information with these apps,” he said. “It’s a great example of what can go wrong.”

Rather than hide those details, all of it should be shown to users, Parecki said.  

Austin agreed, and said apps that ask for permission to Gmail should include a more blatant warning over what the user is handing over.

“I’m not on the OAuth hate bandwagon yet. I do see it as valuable,” Austin said. “But there are some risks with it.”

Fortunately, Google was able to quickly foil Wednesday’s attack, and is introducing “anti-abuse systems” to prevent it from happening again. Users who might have been affected can do a Google security checkup to review what apps are connected to their accounts.

The company’s Gmail Android app is also introducing a new security feature to warn users about possible phishing attempts.  

It’s tempting to install apps and assume they’re safe. But users and businesses need to be careful when linking accounts to third-party apps, which might be asking for more access than they need, Cloudlock’s Kaya said. 

“Hackers have a headstart exploiting this attack,” she said. “All companies need to be thinking about this.”