AI-Driven Intelligence for Federal Agencies: What’s Next?

November 2019

Dating back to his days as a CIA executive, Voyager Labs’ Hendrik van der Meulen has witnessed the intelligence world integrate advanced technologies into its daily operations. We recently caught up with him to learn what’s currently on the minds of leading intelligence experts. 

Let’s start off by discussing some of the buzz you’ve been hearing as you talk to government agencies. I know you speak with them frequently, but what’s changed this year?

The notable trend for me is the realization and appreciation by both “practitioners/operators” and decision-makers in government departments and agencies that the ability to analyze unstructured data from open sources can make a tremendous difference in being effective and addressing the most pressing mission priorities and challenges.  We often note that about 80 percent of the data that is available globally is in unstructured form.  That is a nebulous concept, but has critical implications.  It means that there is a lot to know and understand about people, networks, organizations and issues of interest that can be unearthed with the best technologies. By taking advantage of this publicly available information, analysts can counter hostile, criminal, and violent people, groups, and transnational actors.

Federal agencies have plenty of data at their disposal, but finding the right information is often where things get tricky. If you’re an investigator, where do you even start?

We have found that most federal agencies have no lack of leads or cases they are working on.  On the contrary, they appreciate help to find an updated and open source data layer to help inform them about priorities, and to corroborate their findings.  They get their information from a multitude of sensitive sources; what they value is the ability to update, refresh, and bring a “current view” from the open-source data to further their investigations and research, to find new and immediate leads to close cases, or come to highly reliable conclusions about threats and responses.

One issue that seems to have gained traction is applying AI to intelligence gathering. Based on what you’re hearing, where is this need greatest?

There is a growing appreciation in the federal government – and this applies equally to governments outside the US – that most analysts and researchers are being overburdened and spending their precious time and intellectual capital conducting “manual” or rote analytic functions that can be assigned to a machine, to an algorithm.  The mistake some technology companies and experts make is to present it as a suggestion that machine learning and Artificial Intelligence can take the place of an experienced investigator, detective, special operations soldier, or analyst.  We know there is no way to replace “gut instinct” and a feel for the target – that the value we bring is to help those highly trained specialists free their time to focus on higher-order and high-impact information to inform decision-makers as quickly as possible about the relevance of the billions of new data points being generated every day in open sources.

You’ve spoken about OSINT as a force for countering the rise in domestic terrorism. Is there a way to be proactive rather than reactive to these types of threats?

The similarities in the phenomena of foreign and domestic terrorism are well-established by the many governments dealing with them and the expert academics studying them.  International and domestic terrorists and violent extremists are taking full advantage of the open source platforms available to them to propagate their extremist tendencies: recruit new members, justify their positions and actions, and grow as a force in societies.  It is critical to maintain a constant understanding of the development of this phenomenon, to discern its impact and trends, and – using open-source information – to identify those who are willing to cross the line from misguided philosophies to illegal, violent action against innocent people.

Share this post