Online Security: It’s Time to Upgrade Your Passwords

— -- If you’re protecting important personal accounts with nothing more than a few security questions, it may be time for an upgrade.

A study from Google recently compared the use of security questions to other account recovery methods like SMS (short text messages) and an alternative email address. It found that true or made-up responses to questions like, “What was your mother’s maiden name?” or specific answers to prompts such as “Favorite Superhero,” “First Car” or “Childhood Phone Number,” weren’t a reliable way to recover or protect an account.

The basic takeaway is simple: Account recovery (or account takeover) is easier to accomplish when using “security questions” because too often answers to those questions can be guessed and/or figured out through a process of elimination. The other methods, like SMS or email recovery, require a would-be scammer to have a user’s phone or know how to access the alternate email account used to receive the recovery link. While the study was Google-centric, the same goes for any site you visit that allows users to answer security questions in order to regain access to an account when they forgot their user name or password.

I’ve advocated a form of strategic lying as a way to foil a would-be account take-over, which I explain in my book Swiped: How to Protect Yourself in a World Full of Scammers, Phishers, and Identity Thieves. While I still think it’s a good strategy, the problem with this tactic, as highlighted by the Google study, is the quality of lies told. The most important issue is memorability.

The Google study suggests that most people aren’t very good at remembering their lies, at least when conjuring up answers to security questions. The authors saw “a very sharp decline within 1 month of enrollment for [memorability of] all questions. This is particularly true for 'favorite food' and 'childhood best friend' questions which are not necessarily factual and to which users may change their minds or have to choose from among several possibilities at the time of enrollment.”

While I still think strategic lying works, it’s no longer the best strategy. It only works if you can avoid the pitfalls uncovered in the Google study, and that might be a pretty big “if.”

What Were Those Pitfalls?

The study found that 16 percent of security questions had answers “routinely listed publicly in online social-networking profiles.” Live by the share, die by the share.

This held true even if the users had all their privacy settings toggled tight, because would-be hijackers could use something called an inference attack method, whereby sensitive information can be found (or guessed) by surfing the timelines and other social sharing of a user’s friends.

Identity-related crime is often a family (and friends) affair. The study found that partners, family members, friends and even acquaintances had a pretty good shot at guessing their way into your accounts. As noted, “Schechter et al. found in a laboratory study that acquaintances could guess 17 percent of answers correctly in five tries or fewer, confirming similar results from earlier user studies.”

This is how "Nudegate," happened in 2014, when a number of celebrities’ naked photos and videos were leaked. Only in that instance the problem was the guessability of celebrities’ security questions—people who lived in the public eye, had been interviewed exhaustively and had few secrets. Make no mistake, when it comes to identity-related crimes and the people who commit them, we are all celebrities.

Any opinions expressed in this column are solely those of the author.