Online insecurity

  • March 23, 2012

Analysis of largest ever sample of passwords shows most are easy to crack.

Online passwords are so insecure that one per cent can be cracked within 10 guesses, according to the largest ever sample analysis.
The research was carried out by Gates Cambridge scholar Joseph Bonneau and will be presented at a security conference held under the auspices of the Institute of Electrical and Electronics Engineers in May.
Bonneau [2008] was given access to 70 million anonymous passwords through Yahoo! – the biggest sample to date – and, using statistical guessing metrics, trawled them for information, including demographic information and site usage characteristics.
He found that for all demographic groups password security was low, even where people had to register to pay by a debit or credit card. Proactive measures to prompt people to consider more secure passwords did not make any significant difference.
There was some variation, however. Older users tended to have stronger online passwords than their younger counterparts. German and Korean speakers also had passwords which were more difficult to crack, while Indonesian-speaking users’ passwords were the least secure.
Even people who had had their accounts hacked did not opt for passwords which were significantly more secure.
The main finding, however, was that passwords in general only contain between 10 and 20 bits of security against an online or offline attack.
Bonneau, whose research was featured in The Economist, concludes that there is no evidence that people, however motivated, will choose passwords that a capable attacker cannot crack. “This may indicate an underlying problem with passwords that users aren’t willing or able to manage how difficult their passwords are to guess.”

To read his full research paper, click here.

 

Latest News

Security and risk in the 21st century

What are the major security risks in the 21st century and how should we deal with them? Pranav Ganta is part of the 25th anniversary cohort and will be studying […]

The benefits of bilingualism

Professor Napoleon Katsos is from the first cohort of Gates Cambridge Scholars. Minhee Lee is from the 2025 cohort. Napoleon will be Minhee’s supervisor as she explores the multi-layered meanings […]

‘Personality test’ shows how AI chatbots mimic human traits – and how they can be manipulated

Researchers, including first co-author by Gregory Serapio-García [2019], have developed the first scientifically validated ‘personality test’ framework for popular AI chatbots, and have shown that chatbots not only mimic human […]

Can we use AI to colour outside the lines?

Eryk Salvaggio [2025] and Tristan Dot [2022] are both interested in the questions thrown up by AI art, but they come at it from different perspectives.  Before Cambridge, Tristan studied […]