This discussion is heavily influenced by personal observations of our Nation’s law enforcement (LE) professionals, and how Unmanned Aerial Systems (UAS) are continually pushing the limits of our criminal justice system. Even when a LE agency understands all their authorities (and has some capability to perform an enforcement action in the interest of public safety), there have been very few instances where new UAS-related criminal laws were applied when UAS pilots are suspected of acting in an unsafe or suspicious manner. Following this trend, and then expanding the conversation to include all unmanned systems, our criminal justice system will be challenged when conducting either investigations or enforcement actions when responding to crimes involving new autonomous systems.
First, to keep this thought piece manageable, this paper focuses the conversation on the following unmanned systems: aircraft systems (UAS), ground systems (UGS), seaborne systems (USS), and underwater systems (UUS). For simplicity, I will refer to this group of systems, where the unmanned vehicle changes based upon its deployment, as “UxS.” Although there are other elements of autonomy, artificial intelligence, and deployed robotics that could also cause LE concerns, this paper will describe the current state of UxS, some of which are already being used regularly in the United States. Finally, it should be reinforced that all UxS include similar components; the unmanned vehicle, their sensors or payloads, a data link, a control system, support equipment, and (most importantly) a human operator.
This paper will highlight how LE agencies will face several challenges when investigating and enforcing crimes involving unmanned autonomous systems (UxS):
Overall, the complex and rapidly evolving nature of today’s autonomous systems represents the need to identify evolving challenges and establish new LE paradigms to adapt and prepare for when, and how, criminals will eventually employ these technologies.
The typical unmanned system is any electromechanical system that can execute predetermined or preprogrammed tasks with limited human intervention or control, ranging from remote control to pre-programming, to full autonomy. This discussion is limited to those systems that already exist, whether in the prototype or active employment in modern society.
Referring to UxS, these systems can offer new capabilities beyond the human operator. Examples of these autonomous capabilities can be seen today in:
What is more interesting is that technology is becoming more prevalent when multiple vehicles can be controlled (or monitored) simultaneously. A recent Ukrainian wartime operation highlighted how two separate unmanned systems, both aircraft and surface ships, worked together to successfully target a Russian warship. Specifically, a term becoming more commonly noted regarding UAS is the concept of a “swarm.”[i] However, just because a larger number of aircraft are flying at the same time, does this truly represent a swarm? Think more about a large school of fish or the fascinating display (murmuration) within a flock of birds, all moving together with far greater intrinsic coordination than a typical drone light show. There is technology now that allows for planned simultaneous flights of UAS, and we are now at the advent of the capability where a basic UAS swarm is moving to mimic that of nature. Although modern militaries are significantly interested in what they can do with a swarm, there are also efforts underway to support SWAT interests and HazMat companies who are perfecting the use of multiple distinct UxS platforms.
Also, 5G communications promise new speeds, significantly reduced communications latency, and offer worldwide connectivity to the point where 5G can globally connect “smart cities” as well as providing networks for most autonomous vehicles. A great example is where UAS pilots have recently demonstrated their ability to control their aircraft from one location and fly a mission with minimal latency from another city, state, or country. While exciting, these new technological capabilities also represent an interesting new challenge to LE. There is now a real possibility the machine (UxS) under observation may be difficult to correlate with, or possibly nowhere near the human operator.
I am impressed at some of the LE agencies around the US who are leading the way; however, by virtue of teaching UAS courses around our nation, and highlighting our CUAS LE application limitations, it is apparent there are still a majority of street cops and first responders who are just now learning about these complex UxS concepts. Without a common awareness of these technologies and knowledge of their UxS capabilities, there exists a significant possibility most LEOs could be challenged in their basic duties enforcing laws and regulations and prioritizing the safety of citizens and the protection of the public.
For example, in a public safety situation, important questions come to mind: How is this UxS powered? How can I stop what it’s doing? How can I predict what it will do next? Is what I’m observing normal activity - or potentially a deliberate criminal act? Who is a trusted expert in these new technologies?
Automation refers to three different ideas: the human-machine command-and-control relationship; the complexity of the system; and the type of task being automated. Also, the word “autonomy” can refer to the complexity of the machine. Regardless of the human-machine command-and-control relationship, words such as “automatic,” “automated,” and “autonomous” are often used to refer to a spectrum of the complexity of machines; however, these terms are not synonymous. "Automatic" refers to a system that operates without the need for human intervention yet may still require input or direction. "Automated" refers to the use of technology or a process to perform a task without human intervention. "Autonomous" refers to a system or device that can operate independently and make decisions without human input or direction. The most significant difference is the amount of human input required for this system to work … basically, the degree of the operator/programmer is related to the decision “loop.”
Functions, where the human is IN-the-loop, are those that require a positive affirmation from the human operator for the machine to proceed. This is a simple delegation relationship, such as commanding a UAS to hover, then land. ON-the-loop functions are those where the operator need not approve the action beforehand, but the operator retains the ability to veto it before the execution of the machine’s action or abort the action once it has begun. These “human-supervised” delegatory relationships can be more complicated; however, a simple example would be sitting behind the wheel of a semi-autonomous car. OUT-of-the-loop functions are those that can be initiated by either the human operator or the machine, with various rules for their interaction. These can include offering the human a complete or narrowed list of alternatives to choose from based on the machine’s assessment of the situation, suggesting a course of action to the human operator, and initiating action while permitting the human a limited amount of time to affect its behavior. Of note, there now exist UxS functions where the operator is entirely OUT-of-the-loop, insofar as the UxS can initiate and execute without further interaction of a human operator and cannot be vetoed or aborted. Such systems are said to be fully autonomous and are now being tested in the world of weaponized UAS, such as in the Ukraine-Russia conflict.
Autonomous UxS do not categorically represent a threat. It is the intentions of the human operator, and whether a UxS’s acts are delegated or controlled by the bad actor, that the concept of threat applies. Understanding the risks involved (the human threat, the consequences, and related vulnerabilities), and management of those related risk factors, are key for public safety agencies when faced with an unknown UxS.
Regardless of the level of each system’s autonomy, and how many types of UxS are involved, the UxS is not charged with a crime – only the human can be investigated, arrested, and prosecuted. Specifically restated, when there is a potential UxS-related public safety risk, or a reasonable suspicion that a UxS is involved in criminal behavior, this is where most law enforcement agencies could be most challenged.
Currently, there are multiple US companies with the limited legal ability to mitigate an unwanted UAS; however, these capabilities tend to focus on detection and tracking the aircraft. The most difficult task remains to find its pilot. When CUAS technologies claim the capability to identify the location of the UAS controller, they rely on sensing and tracking the electronic communication link between the aircraft and the pilot. However, success becomes much more difficult should the UAS be programmed to operate autonomously. A worst-case UAS scenario is operating without the need for GPS, employing onboard sensors and some level of artificial intelligence guiding its mission. In such instances, even when the technology is lawfully successful in removing the UAS, LE is still not guaranteed to locate the pilot AND establish enough probable cause to believe the pilot intended to commit a crime.
Regardless, for the police agencies of today, there is one point to emphasize in understanding UxS which should logically guide most LE actions. The first and most pertinent question for LEOs is: Who is the human who programmed or commanded the activity?
To establish probable cause that the operator/programmer of any type of UxS violated a US criminal law, there is typically some requirement to establish an element of human intent. In a best-case scenario, LEOs responding to an unsafe or suspicious incident find the operator and document a consensual interview, which is designed to capture their answers to typical “who, what, where, and why” types of questions. Without human admission, proving the intent behind the activities of UxS becomes more difficult. Sometimes other witness statements or forensics might help, but it is usually a human who provides an on-scene account that allows the LEO enough legal authority (somewhere between reasonable suspicion to probable cause) to actually do something – assuming there is a criminal law that applies. There are some criminal laws and statutes being enacted at both the Federal and State/Local level that apply to UAS; however, there are no new laws written for the other autonomous vehicles.
Furthermore, when considering the world of all other UxS, there does not appear to be any non-military detection and mitigation technologies available to US LE agencies. Additionally, the functionality of a Counter-UxS might not even be practical or available (in a non-military setting) for seaborne and underwater systems. In the absence of Counter-UxS technologies, LEOs might be taught to use their own senses to detect, identify, observe, and report what the machine does in order to infer the intent of the operator. However, this is also not typically possible in most seaborne or underwater UxS scenarios. There literally is no capability for any human to “See something, Say something.”
These are important considerations for LEOs because crimes are happening around the world today involving more than just small UAS. Extrapolation beyond these criminal activities, and their demonstrated innovation, brings up the need for more than just buying a new detection system. LE agencies may need to take a step back to categorize what exists in their Areas Of Responsibility (AOR) to protect and patrol, which also includes our most critical infrastructure. Simple gates and guards protecting significant infrastructure are typically no longer effective when considering open access of UxS to airspace, beaches, shorelines, large bodies of water, rivers, parks, highways, and paths. Also, because UxS can readily cross jurisdictional lines and borders, this may increase the difficulty for LE agencies to determine response protocols, asset sharing, and investigation of a suspected crime.
Because of media reporting, many in the public domain are aware of (semi) autonomous cars and driverless taxis in use within the US. Notably, there is almost no media reporting on trucking companies currently operating autonomous small convoys on US highways and roads. Because of this lack of awareness, when there have been accidents and collisions involving these unmanned systems, this initially challenged the response of LE agencies. There are new important considerations for LE: If multiple types of UxS are involved, are there multiple bad actors to be investigated?
While LE agencies may need to develop policies and procedures in dealing with autonomous vehicle incidents, some are working to improve their understanding of the technology through training and research. Currently, there are organizations and initiatives in place such as the National Highway Traffic Safety Administration (NHTSA) and the National Association of City Transportation Officials (NACTO) that are working to address the challenges and provide guidance for law enforcement agencies to improve their preparedness. While our government works through these challenges, our Nation’s first responders are left to work things out in the interim. Look at two recent incidents when San Francisco firefighters had to disable driverless cars that were interfering with a fire scene. In the latest incident, it took two minutes to break the glass of the car and stop the vehicle. In these situations, there was no operator intent to cause harm or disruption; however, imagine with intent, it could have been worse.
Furthermore, the US Coast Guard (USCG) is researching measures to ensure the security of autonomous vessels and to prevent their use for criminal activities. The USCG has issued guidelines on how to properly secure autonomous watercraft and has set up a task force to study the security implications of USS and UUS. Regardless of the Federal Government’s outreach efforts, there are still places within the US where the “regular street cop” may currently be unprepared for investigations and enforcement actions involving crimes committed with UxS in water-based or mass transportation scenarios. For consideration, recall a July 2019 incident where USCG boarding officers dramatically leaped aboard a semi-submersed “narco-sub” carrying 17,000 lbs of cocaine. It took considerable risk and valor to cause the operator to finally open the hatch and stop the submarine. Should some waterborne UxS be suspected of criminal activity on a local lake, stream, dam, or shoreline, what are the LE agency’s jurisdictional capacities and/or capabilities to stop and investigate who is involved?
As referenced earlier, when an unsafe or suspicious situation involving UxS has been resolved and LEOs have some luxury of time, there is still the requirement to document and present to the criminal justice system what happened. More simply stated, before any case goes to trial, there is an absolute need for evidence. Preserving and collecting all physical evidence from UxS typically requires forensic tools and some training. Some UxS capture and record so much data that specialized equipment and capabilities are necessary to extract relevant information. In fact, most UxS collect vast amounts of data, which could be relevant to an investigation; however, such data may be stored in multiple locations (i.e., within the vehicle, at the control station, on a mobile device, etc.), that it may become extremely difficult to retrieve, safeguard, and analyze. Also, some autonomous systems often store sensitive information, and the possibility of data breaches or unauthorized access must be considered when conducting investigations or enforcement actions.
More and more, some state, local and federal agencies (FBI & DHS) have the capabilities to assist in capturing forensic UAS-related evidence, but this requires expertise and court-approved techniques. As with most “electronic” devices, UAS use a variety of new technologies to allow its proper operation. Technologies range from GPS signaling to RF communications, using external SD card readers, internal memory chips, or electronic flight logs. Much information can be obtained by digital forensic analysis of the UAS, the controller, the desktop app, the mobile app, or even the cloud. If an agency has CUAS detection technology, they may be able to retrieve other pertinent data (location, RF data, flight profile, distance, speed, altitude, etc.) and even the Media Access Control (MAC) address for the UAS operating system. However, is the LE agency, and the prosecutor, comfortable with introducing these concepts at trial?
As an example, a UAS is detected flying suspiciously over an NFL game, yet LE was unable to identify the operator before it departed the area. Later, security footage from the stadium shows the suspect getting into a car. After running the license plate, law enforcement arrives at the suspect’s house and locates a UAS. How do they prove it was the UAS seen flying over the stadium? If a passive RF detection system detected the UAS and recorded its MAC address, it could be compared to the UAS’ MAC address at the suspect’s house. Will the prosecutor’s office be comfortable meeting search warrant requirements to obtain access to the suspect’s computers and flight logs to gather more evidence? Therefore, because the goal within investigations is to prove the UxS operator intended to commit a crime, LE agencies must establish a protocol to capture, preserve, and recreate these types of digital evidence for presentation at a criminal trial. Challenges to consider for LE agencies include: How was the data captured and preserved; What was the probable cause for a search warrant; Is the LE agency comfortable with disclosing their techniques in collecting these data; Could there be other data not captured which could be considered exculpatory, or potentially challenged in court?
Current research has not noted any U.S. arrests correlated to criminal-related use of an unmanned system - with the exception of UAS. In the U.S., there do exist several federal laws and regulations related to UAS which cover issues such as registration, operation, and safety. Notably, because of the relative ease of use and availability of these aircraft, criminals worldwide have exploited UAS capabilities to break laws that are only limited by the operator’s imagination. Unfortunately, just because a lawful arrest is made pursuant to a reasonable belief that a crime is committed, there is no guarantee the criminal will actually be sentenced to time behind bars. This is not a criticism of the capable and interested prosecutors but rather a factor of minimal prior case law, inconsistent judicial interpretation, and sentencing guidelines seem (at times) to not match the potential severity of the crime.
For example, starting in November 2017, Tracy Mapes of Sacramento, CA used his modified UAS to drop anti-government fliers over two NFL games, one state football stadium, and multiple other public events. Up to his trial sentencing, Mapes continued to show no remorse and made statements he would do it again, yet only received 24 months probation. Beginning in 2017, Jason Muzzicato used multiple UAS, modified to drop nails, ball bearings, and variations of modified explosives targeting his ex-girlfriend in Northampton County, PA. Following an extensive investigation, arrest, and search warrants, Muzzicato ultimately pled guilty to possession of firearms by a person subject to a domestic violence protective order, possession of a destructive device, and knowingly operating an unregistered aircraft. He was sentenced to five years in prison and three years of supervised release. Although this is a significant sentence, his only UAS-related crimes included his guilty plea for “unlawful operation of an unregistered aircraft” (a civil fine), with no enhancement due to explosives and no charges related to the FAA’s Reauthorization Act of 2018, Section 363, “Prohibition regarding weapons.”
In a February 2018 US Department of Transportation (DOT) Office of Inspector General (OIG) report, the IG’s office testified, “Since 2016, our Office of Investigations has opened 23 cases involving the illegal operation of UAS. However, 10 of these cases were closed in the preliminary complaint phase, and 9 were declined for prosecution for various reasons, such as the inability to prove criminal intent and a lack of prior prosecutions.” The only type of criminal activity which appears to have consistent federal prosecutions and sentencings (between 12 - 48 months incarceration) is correlated with criminals using UAS to deliver contraband to prisons.
There have also been disturbing instances of criminal use of UAS in other countries. Japan’s Aum Shinrikyo was one of the first terrorist groups to experiment with the nefarious use of UAS. As early as 1993, this religious cult was conducting experiments that would allow them to transport toxic chemicals by remote-controlled helicopters on an unsuspecting public. In August 2002, the Columbian military found nine UAS on a paramilitary base belonging to the terrorist group the Revolutionary Armed Forces of Colombia (FARC) which planned to use the remote-controlled UAS to drop explosives on government targets. In May 2020, a lone-wolf terrorist planned a UAS attack at a soccer game between Barcelona and Real Madrid using a UAS with an IED payload; however, due to the coronavirus shutdown, his plan was foiled. Some notable events in Mexico include many years of smuggling, narco-terrorism, and explosive targeting of police agencies. More alarming is a documented June 2022 use of a DJI AGRAS UAS sprayer</a > which intentionally sprayed two liters of chemicals mixed with urine and feces over a political rally in Brazil.
Currently, there are no federal laws specific to driverless cars in the US. However, the National Highway Traffic Safety Administration (NHTSA) has issued guidelines for the testing and deployment of autonomous vehicles, and several states have passed laws regulating the operation of driverless cars on public roads. These state laws vary in their specifics but generally cover areas such as registration and licensing, insurance requirements, and testing and deployment. Also, there are no specific US federal or state laws that address the operation of fully autonomous surface vessels, boats, or submarines. However, there are laws and regulations that apply to maritime transportation and navigation which could logically apply to autonomous vessels. The USCG is responsible for enforcing federal laws and regulations related to maritime transportation and navigation, and the National Oceanic and Atmospheric Administration (NOAA) has issued guidelines for the operation of autonomous vessels.
With minimal criminal laws in place, and predominately only Federal regulatory oversight, the concern is that there may not be much opportunity for LEOs to apply a law that fits within their jurisdiction or training. This limitation is similar to the case where the Federal Aviation Administration (FAA) has the primary responsibility for regulating the operation of UAS in the United States; however, they typically only enforce violations with non-criminal fines. Because LEOs cannot enforce FAA regulations, they may only serve as trained collectors of evidence and interviews. Following this trend of lack of enforceable laws and inconsistent sentencing, it is logical to extrapolate there exists less of a chance of a successful prosecution when any other type of UxS is used in the commission of a crime.
A nefarious or criminal actor will use anything within their access to accomplish their goals. Drawing from the subset of all unmanned systems, such actors have used UAS as an available tool in their motivations. Specifically, by looking at past intentional misuse of UAS, one can readily forecast future opportunities for deliberate acts using new unmanned systems technologies. Some think tanks and criminology groups are beginning to ponder these implications. Moreover, there are others who have postulated the beyond the UxS machines in this discussion and are bringing up AI-enabled future crime.
One manner of forecasting such behavior is recognizing the past acts of criminals. To that end, LE agencies should reference modern-day international and domestic incidents of intentional criminal and terrorist acts and use current UAS trends as a predictor for future risks. One important note, looking at the intentional malicious abuse of any unmanned system, and the typical motivations of any criminal actor, the manifestation of their intent to misuse emerging technologies will necessitate a review of LE preparedness.
Finally, an unknown or newly recognized UxS is not a definitive risk, per se. Overall, the first step for LEOs in almost any situation would be to identify the operator of these UxS. Only then would the LEO be better positioned to identify the reason for the UxS’s actions, its intended purpose, and the opportunity to apply any available laws. Learning and understanding these new technologies, and leveraging current knowledge of each agency’s AOR, are the best ways to prepare, prevent, and plan if or when the presence of an unsafe or suspicious UxS requires any type of LE response. The frequently quoted adage, “If you fail to prepare you are preparing to fail” forecasts where US lawmakers and LE efforts should turn their focus. The capabilities and permutations of future criminal misuse of UxS will happen someday soon; therefore, LE awareness and planning for these issues are already overdue.
Dan Delgado is Program Technical Lead for AVIAN's Counter Threat Solutions capability. Dan has over 20 years of FBI experience with a demonstrated history of working in Crisis Management, Risk Management, Weapons of Mass Destruction, Counterterrorism, Human Intelligence, and Emergency Management. Dan holds a Master's Degree in Criminology, Weapons of Mass Destruction from the Indiana University of Pennsylvania and a Bachelor's Degree in Civil Engineering from the US Coast Guard Academy.