Mature Your Threat Intelligence Program: How to Measure, Benchmark, and Improve Intel Fidelity

When you think of threat intelligence, what comes to mind? Collections of IP addresses, hashes, maybe domains? Like many organizations, you may have many intelligence feeds with thousands of these indicators. How do you make sense of it? How do you create actionable decisions based on these? How do you measure the fidelity of your intelligence?

To begin to assess your threat intelligence, we need to understand what threat intelligence is and what comprises high-fidelity intelligence.

Data vs. Intelligence

Data is not intelligence. Raw data feeds, logs, IP lists, etc. are not threat intelligence, although they are a part of it. For data to become threat intelligence, it needs to be processed and have context added to it. This gives organizations the ability to make inferences and find correlations in their data to make business decisions related to risk.

Tactical vs. Strategic

Another aspect of threat intelligence is tactical vs. strategic intelligence.

Tactical intelligence is information on what is being attacked and how it is being attacked. An example would be web server X is being attacked with vulnerability Y. This type of intelligence focuses on the specific TTPs of the actor. Typically, this is useful for those directly responsible for defending an organization, such as incident response teams.

Strategic intelligence is the who and the why. This intelligence is usually more a broad overview of a threat landscape. This intelligence helps an organization understand how their organization may be targeted and why. Typically, strategic intelligence is useful for making high-level executive decisions in line with current trends.

Examples:

Tactical:

• My web server is being exploited by a new CVE-XXXX-XXXX

Strategic

• APT28 is attacking my webserver because they want to compromise the government entity hosting the server

Signatures vs. Indicators

Many times, threat feeds will give you indicators and not signatures. While many organizations may treat these as the same, indicators of compromise are not signatures. An indicator is a signal that some type of activity may have happened (IP address, domain, etc). A signature is a cause and effect correlation of a few different indicators that means a suspicious activity likely occurred.

Examples:

Indicators

• An IP Address
• A Hash
• A Domain

Signature

• Outbound traffic with characteristic X, to IP address Y, indicates activity related to actor Z
• Malware strain X always causes outbound traffic with characteristics Y to destinations Z

Using indicators as signatures can in turn lead to higher false positive rates. For example, an IP address may be a load balancer or content delivery network comprised of multiple domains. A domain may be a legitimate service, such as Dropbox. It cannot be assumed the whole Dropbox domain is breached. Ideally, alerting on threat intelligence should be focused more on the high confidence signatures.

Measuring Strength and Fidelity of Indicators

The above graphic is the “Pyramid of Pain” first coined by David Bianco. The pyramid is a ranking system for threat intelligence indicators/signatures. The bottom of the pyramid is the easiest indicators, not only to detect, but also for an adversary to change. As you get closer to the top of the pyramid, these are the indicators that are harder to detect, but also harder for an adversary to change. The higher you are able to go, the more fidelity your intelligence has and the better you can defend yourself against attacks. Let’s briefly break it down.

Hashes

While hashes are a very definitive indicator, (if you see a known bad hash, you can be sure that file is malicious) they are also very “trivial” indicators because how easy they are to change. As an attacker, if I were to change a single bit or add a single null byte, the entire hash is now different and not in a threat intelligence list. If you only look at a suspicious file for its hash and determine if it is safe only based on the hash not coming back as known bad, this can create false negatives. While they are great to have in incident response and threat feeds, they cannot be used as a gold standard.

IP Addresses

IP addresses are arguably one of the indicators people most traditionally think of when they think of threat intelligence (that and domains). Almost all attacks on a network will involve some sort of external IP address. Again, the problem with relying on a list of IP addresses is that by the time they appear on a threat list, there is a good chance they are no longer being used for the malicious purpose that put them on that list. IP addresses can easily be switched by changing infrastructure, VPNs, TOR, proxies, etc.

Domain Names

Domain names are only slightly harder than IP addresses for an adversary to change, as they typically require payment, must be registered, and are hosted somewhere on the internet. Other than these points, they are nearly as easy to change as IP addresses.

Network & Host Artifacts

Going farther up the pyramid, network and host artifacts is where we start seeing both higher fidelity threat intelligence, and where it starts impacting an adversary’s campaigns. These could be user strings, drop locations, registry keys, etc. While they can still be changed, it would typically require updates to tools, research into how it was detected, or other actual effort on behalf of the attacking party.

For threat intelligence, this is where we can start seeing the benefits of correlating our indicators and creating signatures. For example, you detect a host reaching out to malicious IP addresses, so you block the IPs. Soon after, you see the same behavior, but with new IPs. After examining the traffic of each instance, you notice the same suspicious HTTP header used to exfiltrate data. By detecting and blocking this artifact, you not only have higher fidelity intel, you force the adversary to change their tactics.

Tools

For an adversary, they are likely using in-house tools that are not open source, as they will not have public signatures. These tools typically take a significant number of hours to create and have significant financial cost associated with them. To create new tools or learn how to use another existing tool will take significantly more time and effort than changing something like their IP address.

Tactics, Techniques, and Procedures

At the top of the pyramid, we have TTPs. At this stage in the pyramid, your threat intelligence and detections are looking at behaviors and signatures, not individual indicators. This provides visibility into the attack at its core. By discovering and blocking TTPs, you force an adversary to create all new behaviors to continue their attack.

Bringing it Together

To bring it all together, let’s take an example of credential dumping via LSASS, which is then exfiltrated back to the adversary, and compare it to our pyramid.

Hashes

• You block the Mimikatz hash
• The adversary makes custom version

IP addresses/Domains

• You block known bad IPs/Domains being used for exfiltration
• The adversary changes either or both, exfiltrates data

Network/Host Artifacts

• You alert on dump files to the temp folders where the attacker is known to put data
• The adversary changes location, encrypts it, renames it, etc. Although, it likely requires recompiling tools.

Tools

• You discover how the adversary’s custom Mimikatz module behaviors, how it spawns processes, injects, etc. and alert on these new behaviors
• Adversary can change these, but it takes significant effort, and requires more than changing a hash

TTPs

• You alert and block ANY unknown process from dumping LSASS
• The adversary will have to develop a whole new tactic to gain the same credential material

In conclusion, mature threat intelligence is more than just raw data feeds. To get the most out of your threat intelligence , you need to add context and tailor it specifically to your organization’s threat risk. By comparing your intelligence artifacts to a benchmark, such as the pyramid above, you can measure the effectiveness of each and make judgements about your intelligence posture. Try to evolve your intelligence programs to be able to detect and alert on behaviors and signatures, rather than only indicators themselves. By doing these, you will not only give your organization the ability to defend against more sophisticated adversaries, but also give your team the ability to infer, correlate findings, and make business decisions based on your intelligence.

How ReliaQuest GreyMatter Integrates Multi-Feed Threat Intelligence for Comprehensive Coverage

ReliaQuest GreyMatter automatically collects, normalizes, and prioritizes threat intelligence in a consumable format for your SIEM and EDR. ReliaQuest GreyMatter processes all IoCs and only sends those with the highest fidelity, so your security controls report less false positives. Customers on average receive over 35,000 new IoCs each week, ensuring up-to-date, relevant intel for comprehensive threat coverage and a 25% average increase in true positives.