Using Artificial Intelligence to Prevent Lone Wolf Attacks

July 13, 2022

Violence against civilians is always disturbing and disruptive. This is what gives terrorism its power over a population, as unpredictable aggression threatens a society’s fundamental sense of calm and normalcy. For much of the 21st Century, the public fear focused on organized terrorism, planned attacks by centralized groups with top-down recruitment and operations. Global agencies developed strategies for countering these organizations, and they were often effective.

However, as social media and other technologies expanded, terrorist groups became more diffuse, and individuals around the world became radicalized. Groups like ISIL, and many smaller political organizations, loosened their control to encourage anyone who aligned with their ideologies to commit violence in their name. Since then, there has been a pattern of growing autonomy among these criminals, and fewer ties to specific radical groups. The Lone Wolf phenomenon brings several complications for law enforcement and the intelligence community, including:

  • No direct command structure, so their communications are not as easily linked to known radical groups
  • Their ideologies and long-term goals are often undocumented
  • No reliance on organizations to supply weapons / equipment / documentation
  • Advantages as “home-grown” natives who do not require documentation to access their targets
  • No need to bring weapons or other contraband over borders
  • Blending into the targeted populations
  • Intimate knowledge of the areas, targets, schedules, customs, and traditions of the populations they are targeting

The Growing Threat of Lone Wolf Attacks

Lone Wolf attacks continue to escalate around the world. In fact, there have already been over 300 mass shootings in the United States so far in 2022. The precise patterns of motivations can be difficult to pin down, as the assailants may or not have ties to a particular organization or cause. In many cases, investigators find that mental health issues were a driving factor. For instance, the attacker who struck in Norway in late June, 2022, appears to have been ideologically motivated, although it is likely that his mental health issues played a part in his motives. Denmark recently had its first terror attack since 2015 at a mall in Copenhagen, resulting in three deaths and four injuries as the shooter targeted people randomly.

Lone Wolf Attacks are Not Entirely Unpredictable

While Lone Wolf attacks are usually more difficult to forecast than large-scale plots by centralized terror networks, they are not impossible to detect before they happen. In fact, a major reason for the failure of agencies to prevent such tragedies comes down to the lack of applied Artificial Intelligence (AI) analysis. In many cases, the information that is needed to stop an impending attack is available through Open-Source Intelligence (OSINT).

Social Media Analysis

The recent mass shooting in Chicago at a 4th of July celebration provides a sobering reminder that such attacks rarely happen without any warning at all. Robert Crimo had a history of mental illness and a robust social media life that included frequent posts of violent statements and images. AI technology is capable of data mining OSINT information through virtually every available platform in the Worldwide Web, Deep Web, and Dark Web. It can do this in up to 100 different languages, with both linguistic analysis and image analysis.

Linguistic analysis can be programmed to identify complex factors of vocabulary and phrases (including slang and codes) to indicate illegal activities, as well as hate speech, threats of violence, and quotes or references to materials from known extremist groups. In many cases, these individuals are in direct communication with radical factions. Laws regarding this type of expression vary by country, but awareness of such patterns can identify persons of interest.

Imagery analysis, especially when combined with linguistic analysis, is a powerful tool in forecasting violent behavior. This type of AI identifies, matches, and analyzes thousands of images, and can be programmed to find depictions of weapons, explosives, vehicles, locations, documents, and other significant items. When matched to individuals through facial recognition, such technology can raise flags to alert the analysts to the type and source of potential illegal activities and intentions. These systems can be automated to mine and analyze data around the clock, alerting investigators to potential suspects.

In addition to OSINT sources, investigators may have endless options for accessing potentially classified or confidential sources with AI technologies. Some AI systems can coordinate Open-Source data with information from law enforcement databases, mental health agencies, and other records that can establish patterns of violence and mental instability. They may also access records of weapon and equipment purchases, travel, and related activities, all of which can be used to identify those who would bring terror and death to civil society.

Share this post