Google has released research that illustrates the wisdom of its recent decision to drop security questions as a fallback option for forgotten passwords.
What was the name of your first pet? What’s your mother’s maiden name? Everyone has answered these questions when signing up to an online account, which serve as a last resort when people forget their passwords. New research from Google shows that website operators should ditch them, as they’re too easy for hackers to abuse and often not that helpful for users.
The research was conducted by Google researchers, and Stanford University researcher Josep Bonneau, who used “hundreds of millions of secret answers and millions of account recovery claims” from Google to demonstrate that security questions stink in practice as an account recovery method.
For English-speaking users, an attacker would, in a single attempt, have a one in five chance of guessing the answer to the question “What is your favourite food?” The answer apparently is pizza.
They also found that 37 percent of people provide false answers to question under the belief this will make them hard to guess. This may however have a perverse effect on the user’s security since many people choose the same false answers, according to Google, thus increasing the chances an attacker can guess it.
Another issue is that security questions are a terrible way to for users to regain access because difficult questions and answers are hard to use.
According to the research, 40 percent of Google’s English-speaking US users couldn’t remember the answer to their question when they needed to. Though not exactly a equal comparison, Google notes that the recall rate was no less than 75 percent for email and SMS delivered codes. Still, in practice, they’re more efficient for the site operator.
Easy questions — like “what’s your father’s middle name” — have a high recall rate but that comes at the expense of them being relatively easy to guess. An attacker has a 20 percent chance of guessing the right answer within 10 attempts for Spanish-speaking users. A more secure question would ask what a person’s frequent flyer number is, but the recall rate here is nine percent.
The researchers also compared the security of user-chosen secrets with a number of plaintext password breaches, such as RockYou, Yahoo and an iPhone 4-digit PIN leak. They found that questions that ask for a name are roughly as guessable as a 4-digit PIN, while numerical-focussed question were equivalent user-chosen passwords.Read more: How Can I Protect Myself from Identity Theft Online?
Security questions are just one more way that hackers can gain access to an account, alongside key logging malware, and phishing emails and websites.
Google in 2013 stopped collecting personal knowledge questions during web account signup and in September last year eschewed security questions for a secondary email or phone number to recover an account. However, many others, including Microsoft, Yahoo and Facebook still support security questions as a forgotten password fallback. It urged site owners to “think twice” about using secret questions, given its findings.
The research follows another quiet change to authentication by Google that will soon see it split the Gmail login process over two pages. The two-page login hasn’t rolled out globally yet but as it does, Google will also be introducing new authentication methods. The company hasn’t yet disclosed what they will be.
This article is brought to you by Enex TestLab, content directors for CSO Australia.
- Governments reaping security dividends from early action on virtual desktops
- The week in security: It's hack or be hacked as airplane rises, defences fall
- Surveillance laws driving companies to limit data collection, developers to boost security
- Facebook trials end-to-end encryption for email to end-users
- Security skills deficit even hits ACCS as momentum builds industry, research collaborations
- A Quick-fire Guide to Secure Code Development