A Deep Dive on How to Catch Phish
The modern email threat. The simple plain text email appearing to come from the CEO asking the junior finance or accounts payable team member to immediately settle the overdue invoice from an irate supplier, that has just called them personally to complain.
Call it Business Email Compromise (BEC) or CEO Fraud, it’s still a targeted phishing attack, and the number of incidents has been rising steadily. Trend analysis here at CensorNet shows that these emails will soon account for 1% of all emails processed – or 1 in every 100 messages our customers receive.
Defending against this particular threat continues to be a major focus for the team, and an area of significant innovation and investment.
Whilst FBI Operation WireWire resulted in the arrest of 74 individuals in multiple countries last week – that still leaves plenty more Phish in the sea.
The problem with CEO fraud email messages is that they are notoriously difficult to detect.
In a recent attack, the only attribute of a message that was changed was the ‘Header From’ field. The display name in Outlook (other email clients are available) showed the CEO’s name.
(Note: Even the From address in < > next to the display name showed something similar to this email address – email@example.com – which should have been enough to alert the user, but security education is not the topic of this blog post).
Nothing about the sender or sending server was suspicious. The IP address was not in any blacklist, the MX record was valid, the sending server matched domain and responded to an smtp probe. There was no SPF record.
We’re still undecided as to whether this makes the attacker super-smart or simple-stupid. The simplicity of the attack meant the message was likely to make it through most email defences, but would rely heavily on the recipient user being half asleep.
Secure your entire organisation from known, unknown and emerging threats - including email fraud.
What this example does provide, is crystal clear evidence of the need for an ultra-modern and multi-layered approach to email security.
Traditional pattern matching / recurrent pattern matching technology is as much use as a chocolate teapot.
Content analysis – looking for message content that includes ‘urgent wire transfer’ or similar language can be effective but comes at a price. And that price is a risk of false positives – incorrectly identifying legitimate emails as ‘Suspect’.
Although, you could argue that quarantining the occasional message chasing payment of an invoice will help cash flow and is still better than inadvertently transferring $25,000 to an account in China or Hong Kong.
Algorithmic analysis is a powerful weapon in the arsenal for identifying scam emails, but even with over 1,000 algorithms examining over 130 elements of the message (in less than 200ms, about half the time it takes to blink), there was little (read nothing) to fire on in this case.
What was interesting about this particular attack was the domain that was used. It wasn’t a recently registered or new domain – it was almost a month old. It wasn’t a nearby domain (or cousin or typosquatting domain), so Levenshtein distance (one of our favourite algorithms due to its power and simplicity) wasn’t helpful. But. The registrant had a history of criminal activity – registering domains and using them in attacks – and that meant a high threat intelligence risk score.
What the attack also highlights is the need to identify the real names of key individuals in external emails – particularly in ‘Header From’. Building a list of names of the executive team and board members, and anyone else that’s an active spokesperson for the organization, and quarantining messages that contain those names, might not be sophisticated but is still a very valid defence.
As a last resort, some email security solutions rely on the user entering in to a conversation with the attacker – asking for more details about the outstanding invoice, or exactly what detailed (confidential or personal) information the sender needed – building up a risk score with each message exchange until a threshold is reached.
At Censornet we go all-out to invest in combining technologies and techniques that identify and block the initial inbound email. Tracking smtp conversations is still interesting. If a user receives an email from a sender for the first time that also contains potentially suspicious content, then a banner across the top of the email advising caution might just be enough to cause them to stop and think!
Ultimately a combination of content analysis, threat intelligence and executive name checking would have stopped this super-smart, simple-stupid attack. Is it time to think differently about email security – we think so.
Ultra-modern, multi-layered defence wins again.