As we covered in the introduction to our Selecting Enterprise Email Security series, even after over a decade of trying to address the issue, email-borne attacks are still a scourge on pretty much every enterprise. That doesn’t mean the industry hasn’t made progress – it’s just that between new attacker tactics and the eternal fallibility of humans clicking on things, we’re arguably in about the same place we’ve been all along.
As you are considering upgrading technologies to address these email threats, let’s focus on detection – the cornerstone of any email security strategy. To improve detection we need to address issues on multiple fronts. First we’ll look at threat research, which is critical to identify attacker tactics and maintain information sources of known malicious activity. Then you need to ensure detection will scale to your needs, as well as implement some attack specific detection in case of phishing and Business Email Compromise (BEC). Finally we’ll evaluate use of internal email analysis as another mechanism to identify malicious activity within the environment.
Threat Research: the Foundation of Detection
The general tactics used to detect email attacks, such as behavioral analysis and file-based antivirus, are commoditized. There is little value in these tactics themselves, but many detection techniques working together can be highly effective. It’s a bit like mixing a cocktail. You can have five different liquors, but knowing the proportions of each liquor to use lets you concoct tasty cocktails. Modern detection is largely about knowing what tactics and techniques to use, and even more about being able to adapt their composition and mixture because attacks always change.
So threat research is contingent on a mature and robust analytics capability. It’s about blending sources like multiple AV engines, malicious URL databases, and sender reputation databases to determine the optimal mix and weighting of each input. It’s necessary to have a sufficiently large corpus of both good and bad email to identify common components and patterns of malicious email, which then filters back into the detection cocktail.
Threat research requires analytics infrastructure and data scientists to run it effectively. During the courting process with potential vendors it’s helpful to understand their threat research capability in terms of resourcing/investment, skills, and output. Sure, having a research team find a new and innovative attack and getting tons of press is laudable, but it doesn’t help you detect malicious email. We recommend you focus on meat and potatoes activity, like how often detections are changed, and how long it takes a new finding to be rolled out to protect all customers.
Applied Threat Research
Once you are comfortable with a potential provider’s threat research foundation, the next area to evaluate is how that information is put to use within a gateway or service. For instance, how do behavioral detections work within the gateway or service?
You’ll want to know how the offering protects URLs. You learned about their URL database above, but what happens when a URL is not in the database? Do they render it in a sandbox? Do they use techniques like URL rewriting and stripping malicious domains from email to protect users from attacks?
Then focus on finding malicious attachments. How are inbound files analyzed? Does the provider have a sandbox service to perform analysis? What is the latency entailed in analyzing a file, and in the meantime is the message held or sent to the user, while the sandbox runs in the background? Will the service convert files to a safe format and deliver that, while maintaining availability of the original?
What about impersonation attacks (often called Business Email Compromise [BEC]), where attackers try to convince employees that a message is legitimate, and to take some unauthorized action (like wiring a ton of money to their bank account)? This is another form of social engineering, but these attacks can be detected by looking for header anomalies and watching for sender spoofing approaches (such as changing the display name and using lookalike domains). Even something simple like marking messages that come from outside your domain can trigger employees to scrutinize messages a bit more carefully before clicking a link or taking action.
And let’s not forget about phishing. Does the provider have a means of tracking phishing campaigns across their customer base? Can they identify phishing sites and help have them taken down? Phishing is old news, but like many email attacks, seems to have a half-life measured in decades.
Finally, how easy is it to categorize users and build appropriate policies for the group? For example some groups have legitimate business requirements to get files from external sources (including HR for resumes, Finance for invoices, etc.). But some employee groups shouldn’t get many email attachments at all, or are likely to click links to compromised sites. So managing these policies at enterprise scale makes a big difference in the effectiveness of detection. We’ll discuss this more in our next post.
Internal Analysis to Detect Proliferation
Historically email security happened upon receipt of email. Once it was deemed legitimate, a message went on its way to the user, and if the gateway missed an attack you hoped to detect it using another control. Over the past few years more enterprises have started evaluating internal email traffic to detect missed attacks (those dreaded false negatives). For example you can identify lateral movement of an attack campaign by tracking the same email to multiple employees.
The ability to monitor and even remove malicious emails from a user’s mailbox can offer a measure of retrospective protection, addressing the fact that you will miss some attacks. But once you identify a message as bad, you can find out which users received it, how many opened it, and whether they clicked the link – and remove it from their inboxes before more damage occurs.
Another advantage of integrating security with internal email servers is outbound protection. You can check email for sensitive data and malicious attachments before it is sent, providing an earlier chance to stop an attack than having an MTA or egress filter inspect messages on their way out.
Optimally you should detect and block every malicious email, but in the real world the ability to take action after the fact provides more flexibility to protect users.
Sharing Threat Intelligence
One last critical capability to evaluate is how threat intel from your email security vendor can make other security controls more effective. To the degree that sender reputation or attack patterns are enumerated in machine data, other security devices and services can consume that intelligence directly. For instance email threat intel could be loaded into your SIEM to look for network traffic to known spam or phishing domains. Likewise, those addresses could be used to block outbound traffic within your egress filters.
As discussed we have made progress in detecting recent email attacks. But evaluating potential vendors against modern techniques increases your ability to protect your organization’s email effectively. In our next post we’ll dig into how to scale email security to your enterprise, including considering a service versus an on-premise gateway (or both), direct integration with other enterprise security controls, and complimentary services, which can improve your entire security posture when used together.