Dark Reading is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.


02:00 PM
Mike Convertino
Mike Convertino
Connect Directly
E-Mail vvv

Battle of the AIs: Don't Build a Better Box, Put Your Box in a Better Loop

Powered by big data and machine learning, next-gen attacks will include perpetual waves of malware, phishes, and false websites nearly indistinguishable from the real things. Here's how to prepare.

We've heard a lot about the fallout from last year's Shadow Brokers bombshells, particularly when it comes to previously unreleased exploits put into the wild. Those exploits include EternalBlue, which was weaponized into the WannaCry ransomware that wreaked havoc on the Internet.

However, the most damaging element wasn't necessarily the actual exploits. Yes, there has been an incredible amount of harm done, but it only foreshadows the real damage control the industry will be doing in the coming months and years.

Included in the information leaked by the Shadow Brokers, but not as widely understood, was a trove of information on the intelligence community's methodology for finding vulnerabilities and building exploits. Now that this has been exposed, security pros had better be prepared to weather continuous attacks by zero day exploits against any and all applications and platforms.

In the next 12 to 36 months, we're going to see hackers using these techniques to build the next generation of attacks. There will be perpetual storms of malware, click-free attacks, perfect lures — and much of it will be untraceable, with exploits becoming unique and essentially disposable.

Powered by big data, machine learning, and natural language processing engines, we'll see phishes and false websites that will be nearly indistinguishable from the real things. There won't just be new worms, but the equivalent of Web drive-by attacks extended to major services and even mobile platforms.

These attacks will be launched from command and control (C&C) networks that were never seen before and will never be seen again. Expect malware that is constantly evolving and never reused, or which contains so many new detections that traditional antivirus software can't handle it all.

As such, domain analysis for malware C&C networks will become an obsolete art. IP reputation filters will be useless. Information from intelligence providers will become less valuable.

Get Your Head out of the Gear and into the Analytics
To defend against this dizzying new reality, we need to produce new types of logs that are more behavior-based and do a better job of using automation to analyze and detect other outputs across the entire stack. Security professionals need to be thinking about how they're building their fully automated analytical loop instead of what a specific device is detecting.


Learn from the industry’s most knowledgeable CISOs and IT security experts in a setting that is conducive to interaction and conversation. Click for more info and to register.

Essentially, today's gear represents a collection and input mechanism. It's a collector and an actuator. It's both an earphone and a switch. To defend against the attacks of tomorrow, we need to extend the power of that collection effort, and close the loop with machine learning logic that can correlate, corroborate, and take action appropriate to the server or device in question.

To do this, companies must first establish methods for collecting information from all layers. The way to defend goes back to TCP/IP and the Open Systems Interconnection (OSI) model, from the physical layer to the application layer and everything in between.

Improve Your Ability to Detect Anomalies and Close the Loop
There are multiple places to detect anomalies — end-user machines, network gear, firewalls and application-aware firewalls, servers. Output feeds about the behavior of the full stack on these devices must be collected from all these physical locations.

After establishing a collection methodology, it's time to get better at identifying anomalies, with the idea of creating an engine that will know the markers across all layers and devices.

Once those anomalies are understood, they must be fed into some kind of analytical system to be correlated. This allows for corroboration of what's happening at the different layers, enabling more assured detection.

Then there should be some logic that looks at what the anomaly is and where it is best to halt the activity with the least impact to operational processing. At this point, the system can decide where to stop the traffic and make a determination. And the answer may be: multiple places.

In other words, we're going to get to a point where the system doesn't just automatically detect something and corroborate it but goes beyond that to determine the best place to stop it and take action to close the loop.

Don't Be Afraid to Get Crafty
Yes, there's machine learning embedded in this process. Is it something you can build? Yes. Something you can buy? Not always. Notice that I called for the feeds to go to analytical system and not a SIEM. The concept is not fully fleshed out in the industry, though a lot of players are working on it, especially in the SIEM market. The next generation of security analysis capabilities will clean all the disparate inputs, normalize them so that they can be used for analysis, and allow us to compare information from all sources.

In the end, this process will combine and automate network and application security to an extent most organizations haven't experienced before. But to do so, companies will have to get really knowledgeable about what is happening in their network and what the blind spots are. We have to ask questions such as: Do I have anything that gives me visibility into what is happening at the session layer on a PC? To what extent do I have the stack in my PCs, servers, and network covered?

All this is imperative today. In this new era, companies who rely on block lists, human analysis, or end users being able to spot phishing emails are going to be completely exposed.

And the pace is about to change due to our adversaries' ability to generate new attacks in an automatic way. How do we fight off more automation on the bad guys' part? The answer is a massive push into automation and a clearer and speedier corroboration of data on ours.

Related Content:

Mike Convertino has nearly 30 years of experience in providing enterprise-level information security, cloud-grade information systems solutions, and advanced cyber capability development. His professional experience spans security leadership and product development at a wide ... View Full Bio
Comment  | 
Print  | 
More Insights
Newest First  |  Oldest First  |  Threaded View
SOC 2s & Third-Party Assessments: How to Prevent Them from Being Used in a Data Breach Lawsuit
Beth Burgin Waller, Chair, Cybersecurity & Data Privacy Practice , Woods Rogers PLC,  12/5/2019
Cybersecurity Team Holiday Guide: 2019 Gag Gift Edition
Ericka Chickowski, Contributing Writer,  12/2/2019
Register for Dark Reading Newsletters
White Papers
Cartoon Contest
Current Issue
Navigating the Deluge of Security Data
In this Tech Digest, Dark Reading shares the experiences of some top security practitioners as they navigate volumes of security data. We examine some examples of how enterprises can cull this data to find the clues they need.
Flash Poll
Rethinking Enterprise Data Defense
Rethinking Enterprise Data Defense
Frustrated with recurring intrusions and breaches, cybersecurity professionals are questioning some of the industrys conventional wisdom. Heres a look at what theyre thinking about.
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
PUBLISHED: 2019-12-09
alter.c in SQLite through 3.30.1 allows attackers to trigger infinite recursion via certain types of self-referential views in conjunction with ALTER TABLE statements.
PUBLISHED: 2019-12-09
In "Xray Test Management for Jira" prior to version 3.5.5, remote authenticated attackers can cause XSS in the generic field entry point via the Generic Test Definition field of a new Generic Test issue.
PUBLISHED: 2019-12-09
In "Xray Test Management for Jira" prior to version 3.5.5, remote authenticated attackers can cause XSS in the Pre-Condition Summary entry point via the summary field of a Create Pre-Condition action for a new Test Issue.
PUBLISHED: 2019-12-09
radare2 through 4.0.0 lacks validation of the content variable in the function r_asm_pseudo_incbin at libr/asm/asm.c, ultimately leading to an arbitrary write. This allows remote attackers to cause a denial of service (application crash) or possibly have unspecified other impact via crafted input.
PUBLISHED: 2019-12-09
In the macho_parse_file functionality in macho/macho.c of YARA 3.11.0, command_size may be inconsistent with the real size. A specially crafted MachO file can cause an out-of-bounds memory access, resulting in Denial of Service (application crash) or potential code execution.