This article is more than 1 year old

Some credential-stuffing botnets don't care about being noticed any more

They just take a battering ram to the gates

The bots spewing out malicious login attempts by the bucketload appear to have cranked it up a notch.

According to Akamai's latest State of the Internet report on credential stuffing (PDF), its customers alone were deluged by 30 billion malicious logins between November 2017 and June this year, an average of 3.75 billion per month.

This intensified during May and June, when traffic spiked to more than 4 billion malicious login attempts – a bad portent for what might be coming in the rest of 2018.

Credential stuffing is the technique of using a bot to try logins stolen during phishing attacks and data breaches on lots of other sites to see how many succeed. Because bazillions of netizens have the habit of reusing the same password over and over, plenty do succeed.

The method, which took off two years ago, quickly gained popularity as cybercrims leapt upon the lucrative bandwagon.

According to Akamai's senior director of security, Jay Coley, the primary aim of credential-stuffing bots is to test accounts to see whether stolen logins work. If they do, they're sold to other criminals.

The rise of credential bot volume is being driven by its success. "If it wasn't profitable, they wouldn't do it," said Coley. "They see this as like a goldrush."

Volume is a double-edged sword for these miscreants. On the one hand, the bad guys need to try as many logins as possible in an attack, because more attempts equal a greater potential success of breaking into accounts. But ramp up volume too far and the defenders will notice, potentially mistaking traffic for a denial-of-service attack.


Stand up who HASN'T been hit in the Equifax mega-hack – whoa, whoa, sit down everyone


The preferred method was "low and slow", trying to hide malicious logins within the normal traffic volumes. "The clever ones drip feed one or two logins per hour. They come in under the radar," Coley observed.

The report also documented a large credential-stuffing attack where a US credit union noticed malicious login traffic had spiked from an every-day level of 800 per hour to 10 times that volume – 8,723 attempts per hour. Over the week, the union saw 315,000 malicious login attempts from nearly 20,000 different IP addresses. And yet with only 4,382 HTTP User Agent connections from fewer than 2,000 autonomous system numbers, the bot was small by the standards of the mega-bots people are used to hearing about, which might be why most of the active ones haven't been given fancy names yet.

Most of the credential-stuffing bot traffic (2.82 billion attempts) originated in the US, with Russia accounting for a further 1.55 billion. The UK was a very distant sixth with under 200 million.

The US also accounted for the overwhelming majority of the target accounts. This might be partly skewed by the US bias in Akamai's customer base but American criminals have also built larger "dictionaries" of US-oriented usernames and passwords after big breaches.

This implies that the rise in credential-stuffing bots is being fuelled by growth in the number of breached credentials that can be targeted.

Short of imposing authentication and magically abolishing crap passwords, can the bots be stopped?

Whether organisations in the sectors being targeted by these attacks are ready for extra layers of fraud detection is hard to assess, although history isn't encouraging.

It seems more probable – like the bad password habits they have allowed their users to get away with for aeons – they'll put up with a bad situation for as long as possible and invest when it's too late. ®

More about

More about

More about


Send us news

Other stories you might like