AI as a Force Multiplier
- Michael Burgess
- 18 hours ago
- 17 min read
Practical, Realistic Uses of AI
to Support Proactive Crime Prevention
Artificial intelligence is a growing part of modern policing and public safety operations. As its role expands, it must be governed, tested, monitored, and used in ways that protect both public safety and public trust. Current guidance from IACP, NIST, NIJ, and the Council on Criminal Justice reflects that same basic idea: AI may offer real value, but it should be introduced carefully, evaluated honestly, and managed with clear guardrails [1][2][3][4].
There is already significant research, discussion, and institutional guidance emerging around the use of AI in law enforcement and the broader criminal justice system [1][2][5][6]. My purpose here is not to undermine or question existing guidance, nor to persuade either skeptics or early adopters. AI is already a significant and growing topic in law enforcement. Some agencies are embracing it; others are cautious or leery. My goal is simpler: to provide practical, realistic insight into how AI may be used as a force multiplier in ways that support proactive crime prevention, improve public safety, and help agencies think more clearly about what is useful, what is appropriate, and what must be governed carefully [1][2][5].
When I talk about AI in this article, I am referring primarily to the strategic use of AI in law enforcement and public safety operations, including tools and systems that support analysis, prevention, investigations, situational awareness, and more effective use of technology and data. General-purpose AI platforms may also have some practical support value, but they should be understood as only one part of a much broader conversation. NIJ’s AI work highlights areas such as video and image analysis, DNA analysis, gunshot detection, and crime forecasting, while IACP’s materials focus on use cases, policy concerns, and responsible implementation in policing [1][2][6][7].
My goal has not changed. I want to help agencies, officers, supervisors, analysts, investigators, and key partners—regardless of rank, role, agency size, or jurisdiction—think more clearly about what they can realistically do to proactively address, deter, and prevent crime, improve public safety, strengthen communities, and help create safer workplaces for law enforcement as well.
The more useful question
The more useful question is not, “Should law enforcement use AI at all?” It is, “How should law enforcement use AI strategically, thoughtfully, lawfully, and in ways that actually improve prevention and public safety?”
That is the question that matters because AI is not a policing strategy by itself. It is a tool. It can improve speed, scale, pattern recognition, information processing, and triage, but it does not replace evidence-based policing, problem-oriented policing, SARA, focused deterrence, hot spots policing, CPTED, situational crime prevention, or strong police-community-partner relationships. The professional guidance coming out of IACP and NIST supports this problem-first, governance-first approach rather than simple adoption for adoption’s sake [1][3][7].
Agencies should not start with, “How do we use AI?” They should start with, “What problem are we trying to solve?” If the problem is gun violence, auto theft, burglaries, repeat disorder, school safety, overdoses, nuisance properties, retail theft, or recurring victimization, the next question becomes whether AI can help the agency identify patterns sooner, use its data better, prioritize resources more intelligently, and intervene earlier. NIST's AI Risk Management Framework and its field‑testing guidance both emphasize a risk‑based approach: agencies should identify risks, measure performance, and compare AI tools to a clear baseline before relying on them operationally or embedding them in everyday practice [3][8].
In other words, AI is not the strategy. AI helps make sound strategies more practical, more timely, and more scalable.
Why this matters now
This topic matters now because policing cannot stand still while society changes around it. Technology changes. Offender behavior changes. Information environments change. Evidence sources change. Community expectations change. Public safety operations change. Modern policing must adapt too, but it should do so strategically and thoughtfully rather than recklessly. NIJ’s work on AI and place-based policing, along with the National Academies’ proceedings on predictive policing, reflects a profession trying to sort out how advanced analytical tools may contribute to crime prevention while also raising real questions about fairness, trust, and implementation [2][6][9].
At the same time, agencies should not pretend AI is some distant issue that does not affect them. IACP says AI usage is spreading quickly in policing and has built a dedicated resource hub with primers, use cases, and policy guidance [1]. The National Policing Institute has also noted that many smaller and mid-sized agencies are only beginning to explore AI or are already using it indirectly through products and services they have purchased [5].
That matters because many agencies are already encountering AI through tools they use every day or are considering right now. Body-worn camera systems, license plate reader systems, video analytics, digital evidence tools, gunshot detection systems, transcription and translation software, and report-support tools increasingly involve AI or related machine-assisted capabilities [2][6][10]. NIJ explicitly identifies video and image analysis, gunshot detection, DNA analysis, and crime forecasting as major criminal justice AI areas [2][6].
So the issue is not whether law enforcement will encounter AI. The issue is whether agencies will understand it, govern it, test it, and use it in ways that actually support proactive prevention and public safety.
AI is a force multiplier, not a replacement
AI should not replace analysts. It should strengthen them.
It should not replace police knowledge and experience. It should sharpen them.
It should not replace supervision, leadership, legal review, or community partnership. It should support them.
That force-multiplier idea matters because it helps set the record straight. This is not about handing policing over to algorithms. It is about helping agencies process more information, identify patterns sooner, reduce administrative drag, and make better use of limited resources [1][5][6].
For smaller agencies especially, that matters. A department does not need a real-time crime center or a large analytics division to benefit from better information handling. A chief, captain, lieutenant, sergeant, investigator, or designated officer can use AI-supported tools to help summarize weekly calls, identify repeat locations, organize trends, translate information, assist with evidence review, or prepare more useful briefings [5][10].
That does not replace a professional analyst, but it can create analytical lift where there previously was very little.
Start small and build deliberately
For many agencies, the best place to start is not with the most advanced or most controversial use case. It is with smaller, practical uses that improve efficiency, analysis, and prevention without overcomplicating operations. That might include AI-assisted report organization, translation, transcription, digital evidence triage, recurring hot spot analysis, or weekly crime-pattern summaries for supervisors and analysts [1][10]. IACP’s primers and NIJ’s generative AI study both support focusing on practical use cases and adoption factors rather than jumping straight to the highest-risk applications [1][10].
That is often where agencies can get the most honest value. Start with a real problem. Start with a manageable use case. Test it. Learn from it. Adjust. Then decide whether expansion makes sense.
Data quality still matters
AI will only be as useful as the data, inputs, and human understanding behind it. If an agency’s data is weak, incomplete, inconsistent, or poorly understood, AI may simply process bad information faster. That is why data quality, consistent reporting practices, and analyst or supervisory review still matter. Europol’s guidance on AI bias in law enforcement warns that bias and quality problems can emerge across the lifecycle of an AI system, while NIST’s framework emphasizes trustworthiness and risk management in design, development, use, and evaluation [3][11].
That is a practical point agencies should not miss. Better tools do not erase weak data discipline.
Where AI can realistically help proactive crime prevention
The strongest and most defensible uses of AI in policing are often not the most dramatic ones. They are the practical ones. They help agencies detect patterns, identify repeat harm, get more value out of existing systems, improve triage, support more focused deployment, and intervene earlier.
1. Better place-based analysis and prevention
One of the most realistic uses of AI is helping agencies understand where harm clusters and what conditions are associated with that risk. NIJ’s work on place-based policing discusses the move from mapping to forecasting and the value of identifying places at greater risk and the environmental factors contributing to that risk [6][12].
This is where tools like Risk Terrain Modeling fit very well. NIJ describes Risk Terrain Modeling (RTM) as a science‑based method of identifying and measuring crime risk posed by features of physical locations, and as a place‑based forecasting technique that diagnoses spatial risk factors associated with criminal behavior [12][13][14].
That makes RTM a strong practical example here. An agency can use RTM or similar place‑based analytical approaches to identify micro‑locations associated with recurring harm and the environmental conditions driving that risk, then respond more intelligently.
That might mean directed patrol at the right times, changes in lighting, changes in traffic flow, trespass enforcement, landlord engagement, nuisance abatement, school coordination, better camera placement, or business outreach [12][13]. The value is not in claiming that “the computer predicted crime.” The value is in helping the agency understand where risk is concentrating and why, so it can respond more strategically.
A realistic example would be a city identifying repeated late-night violence around a small number of bars, gas stations, and parking lots. AI-assisted analysis or place-based forecasting tools might show that the problem is not spread evenly across the city. It is concentrated at micro-places where crowding, poor lighting, vehicle congestion, and closing-time behavior overlap. That allows police and partners to focus patrol, environmental changes, business expectations, and prevention efforts where they are most likely to matter.
2. Faster recognition of repeat and near-repeat harm
Another strong use of AI is helping agencies recognize repeat and near-repeat patterns faster than manual methods alone. That includes recurring vehicle break-ins around event venues, repeat assaults at bar closing times, recurring retail theft in certain corridors, repeated disorder at one apartment complex, or repeated retaliation risks after violent incidents. IACP’s AI guidance highlights predictive and operational support uses, and NIJ’s place-based work shows why earlier recognition of patterns matters for intervention and prevention [1][6][7].
This is where AI can help detect what would otherwise remain buried in CAD narratives, RMS records, tips, and field observations until the pattern has grown worse. That matters because proactive crime prevention often depends on timing. If an agency can identify a near-repeat burglary series, a recurring theft trend, or a growing problem location early, it has a much better chance to intervene before the next incident rather than simply document it afterward [6][12].
A realistic example would be a suburban department noticing a rise in vehicle larcenies. AI-assisted analysis shows that most of the incidents cluster around school events, gym parking lots, apartment overflow lots, and weekend tournament traffic. That leads to targeted patrol, signage, community alerts, temporary camera placement, and outreach to facility managers. Again, the technology did not solve the problem by itself. It helped the agency see the pattern sooner and act more intelligently.
3. Better use of cameras, BWCs, LPRs, and related tools
AI can also help agencies get more value out of technologies they already use or are already considering. NIJ points to video and image analysis and gunshot detection as major AI-related criminal justice areas [2][6]. Europol’s report on AI and policing identifies computer vision, video monitoring and analysis, large and complex data sets, digital forensics, and strategic planning as important areas where AI may support law enforcement operations [11][15].
In practical terms, this can include faster review of body-worn camera footage, automated transcription and translation, smarter video search, identification of important moments in footage, improved triage of large digital evidence sets, and more effective use of LPR-related information [6][15]. It can also include using camera analytics to help identify unusual crowd movement, suspicious activity, or vehicles of interest in ways that support earlier intervention or more focused follow-up [6][15].
A realistic example would be a downtown corridor with recurring late-night fights. AI-supported camera analytics might help flag crowd surges, fast movement, clustering, or other indicators near closing time so that supervisors and officers can position themselves more effectively. Another example would be an investigator using AI-enhanced review tools to search digital evidence and BWC footage more efficiently across a series of linked incidents. These are force-multiplier uses that improve awareness and speed without replacing human judgment.
4. Better triage of overwhelming information
Many agencies are not suffering from a lack of information. They are overwhelmed by it. CAD entries, RMS reports, tips, complaints, field interviews, partner information, school concerns, social media data, digital downloads, and video all create a large volume of material that is difficult to process quickly. Europol specifically identifies large and complex data sets, OSINT, natural language processing, and digital forensics as major areas where AI can support law enforcement operations [15].
That is where AI can be especially useful as a force multiplier. It can help summarize, organize, compare, extract, and cluster information so that humans can spend more time deciding what matters and what to do next. CCJ’s user decision framework and NIJ’s generative AI landscape study both point toward evaluating real use cases in real operating contexts rather than treating AI as one generic concept [4][10].
A realistic example would be a detective or analyst working a conflict between groups tied to multiple shootings, threats, and retaliatory indicators. AI-supported summarization and extraction tools could help identify recurring names, locations, vehicles, or time patterns across a large number of reports and digital records. That does not replace investigative thinking. It gives the detective or analyst a faster starting point and helps organize the work.
5. Supporting focused deterrence and targeted prevention
The person-focused side of AI requires more caution, but caution does not mean there is no role for data-supported prioritization. The National Academies’ proceedings note serious concerns and limitations associated with person-based predictive policing, including distrust, fear, and implementation challenges [9]. At the same time, those proceedings note evidence of effectiveness for focused deterrence as a distinct violence-reduction strategy that depends on partnerships with community members and service providers [9].
That distinction matters. AI may help agencies organize data that supports focused deterrence, violence reduction, custom notifications, or intervention planning. It may help identify conflict clusters, victim-offender overlap, high-risk groups, or locations associated with serious harm. But the response still needs to be grounded in actual conduct, legal standards, meaningful human review, and legitimate partnerships with prosecutors, probation, schools, outreach providers, social services, and community stakeholders [4][9]. CCJ’s principles and framework reinforce the importance of reliability, accountability, and context-specific evaluation when criminal justice agencies consider AI tools [4][16].
In other words, AI should support targeted, explainable, harm-focused intervention. It should not become an excuse for automated suspicion.
6. Expanding capability across rank, role, and agency size
One reason this topic matters is that it is not just for big-city departments. The National Policing Institute has noted that smaller and mid-sized agencies are often only beginning to explore AI or are using it through purchased solutions and vendor-supported systems [5]. IACP’s resource hub likewise reflects a broad policing audience rather than only large departments with extensive technical infrastructure [1].
That means this conversation applies to chiefs, sheriffs, command staff, patrol supervisors, investigators, crime analysts, school resource officers, community-policing personnel, prosecutors, probation officers, code officials, and other public safety partners.
For a chief or sheriff, AI may support smarter deployment, better accountability, and better use of scarce resources.
For a commander or supervisor, it may support better briefings, better trend recognition, and more informed operational planning.
For an investigator or analyst, it may support better triage, evidence review, pattern recognition, and information organization.
For patrol officers, it may support more focused assignments, better awareness of repeat-harm places, and stronger prevention-oriented deployment.
For key partners, it may support a clearer picture of recurring problems and a more targeted role in prevention.
That is part of what makes this practical. The question is not whether everyone needs to become an AI expert. The question is whether the people doing the work can understand how these tools may help them do the work more effectively and more responsibly.
Not every use carries the same level of risk
This is important to say plainly. Not all uses of AI carry the same level of risk. Using AI to assist with transcription, translation, summarization, scheduling, or pattern detection is different from using it in ways that could influence enforcement decisions, surveillance practices, evidence handling, or charging decisions. The more consequential the use, the more important governance, legal review, testing, auditing, and oversight become. NIST’s field-testing guidance and CCJ’s recent assessment framework both support a risk-based approach rather than treating all AI uses the same [4][8].
That distinction helps agencies think more clearly and start more responsibly.
Policy before deployment, not after
Agencies should not buy or deploy first and write policy later. They should not wait until a tool is already in regular use to begin thinking about policy, legal review, training, oversight, and disclosure questions. Those things should be addressed before the tool becomes part of normal operations [1][3][7]. IACP’s guidance to police leaders, NIST’s risk-management approach, and CCJ’s recent framework all support upfront governance rather than after-the-fact cleanup [3][4][7].
Agencies should also confer with prosecutors and legal counsel before implementing or operationalizing AI in ways that may affect investigations, enforcement decisions, evidence handling, disclosure obligations, privacy concerns, or case preparation. That coordination can help surface legal, constitutional, evidentiary, and policy considerations early, before a tool or workflow becomes embedded in practice [3][4][7][8].
Beware vendor hype
Agencies should be cautious about vendor claims and should not assume a product works as advertised simply because it is marketed as ‘AI‑enabled.’ New tools should be evaluated in the agency's real-world environment against clearly defined needs and a documented baseline before they are trusted operationally. NIST's field testing recommendations are especially useful here because they emphasize baseline comparison and performance testing, while NIJ's generative AI study encourages decision makers to weigh benefits, limitations, and adoption factors before implementation [8][10].
A sales pitch is not the same thing as validated performance.
Define success in operational terms
Agencies should define what success looks like before implementation. In many cases, success is not that the agency “used AI.” Success is fewer repeat calls, faster identification of recurring harm, better evidence triage, more focused deployment, stronger partner coordination, reduced analyst backlog, or measurable reductions in repeat problems. NIST's testing framework and the Council on Criminal Justice’s assessment tools both stress outcome‑based evaluation: agencies should judge AI tools by whether they improve real operational outcomes—such as reduced repeat harm or better triage—rather than by novelty or marketing claims [4][8].
That is a very practical point. The goal is not to say the agency is modern. The goal is to solve real problems better.
Training and buy-in matter
If AI tools are going to be useful in practice, the people expected to use them need to understand what the tool does, what it does not do, when it can help, and where its limitations are. Without training and buy-in, even potentially useful tools may be ignored, misused, or over-relied upon. IACP’s policy guidance, NIJ’s generative AI study, and CCJ’s framework all reinforce the importance of informed implementation and role clarity [4][7][10].
That applies at every level. Leaders need to understand governance. Supervisors need to understand operational use. Analysts and investigators need to understand strengths and weaknesses. Patrol needs to understand what is actually useful in the field and what is not.
How responsible AI use can help communities and partners
A practical article on this topic should not focus only on what AI does for police. It should also explain how responsible use can help communities and key partners.
Used properly, AI can help agencies be more focused and less wasteful. It can help narrow attention to the places, times, and patterns most associated with harm rather than relying on broad assumptions or generalized enforcement. CCJ’s principles explicitly tie AI use in criminal justice to both public safety and individual rights, while stressing public trust [4][16].
That matters for communities. Communities benefit when police responses are more precise, more thoughtful, and more connected to actual recurring harm. Businesses benefit when repeated theft or disorder patterns are identified and addressed earlier. Schools benefit when agencies can recognize recurring safety concerns sooner. Landlords, code officials, prosecutors, probation officers, outreach providers, and community organizations benefit when the operating picture is clearer and the problem-solving effort is more targeted.
In that sense, AI can strengthen the broader public safety ecosystem, not just the police department. It can improve shared awareness, support more targeted collaboration, and help multiple partners focus on prevention rather than simply responding after harm has already occurred.
Transparency matters too
Transparency should be part of the conversation as well. If agencies expect public trust around AI, they should be prepared to explain in plain language what types of AI-enabled tools they are using, the general purpose those tools serve, what policies and safeguards govern their use, and how human review and accountability are built into the process. That does not mean disclosing sensitive operational details or giving away investigative methods. It means being open enough to show that the technology is being used deliberately, lawfully, and responsibly [3][4][17].
In practical terms, that may include a public-facing policy, a general explanation of approved uses, clear internal oversight, and a way for community members to ask questions or raise concerns.
The guardrails matter
None of this means agencies should move forward carelessly. In fact, governance is one of the strongest themes across the current literature and professional guidance.
NIST says its AI Risk Management Framework is meant to help organizations better manage risks to individuals, organizations, and society associated with AI [3].
NIST’s field-testing recommendations for law-enforcement AI tools also stress the need to identify a baseline and measure performance, risks, and benefits against it before relying on an AI tool operationally [8].
IACP says police agencies must be aware of the potential risks associated with AI and make informed decisions about responsible use [7].
Europol’s practical guide on AI bias in law enforcement warns that bias can emerge across the system lifecycle and that over-reliance on AI outputs can lead to harmful operational consequences [11].
CCJ’s AI principles and assessment framework likewise stress reliability, security, fairness, transparency, accountability, and context-specific evaluation [4][16].
Those are not side issues. They are central issues.
Setting the record straight
A balanced discussion on AI in policing should set the record straight on a few things.
First, caution is appropriate. Concerns about privacy, bias, explainability, over-reliance, security, misuse, and trust are legitimate and are reflected in current guidance from NIST, IACP, Europol, and the Council on Criminal Justice [3][7][11][16].
Second, caution is not the same as avoidance. Law enforcement should not simply disregard AI because it raises hard questions. The better approach is to study the research, examine the guidance, define the use case, set the guardrails, test the tool, train the users, monitor performance, and use it in ways that actually improve prevention and public safety [1][3][4][8].
Third, agencies should be transparent enough to build trust, but not so transparent that they undermine legitimate operations. The public should generally know that AI is being used, what broad purpose it serves, and what oversight exists. They do not need every tactical setting or every operational threshold.
That balance is consistent with current trustworthiness, accountability, and transparency guidance from NIST’s AI Risk Management Framework and its companion resources, as well as CCJ’s principles [3][4][16][17].
Bottom line
AI should not replace analysts. It should strengthen them.
It should not replace police knowledge and experience. It should sharpen them.
It should not replace leadership, legal review, supervision, or community partnership. It should support them.
Used strategically, thoughtfully, lawfully, and with proper governance, AI can help agencies identify patterns sooner, use their data better, make smarter use of BWCs, LPRs, cameras, and digital evidence, support more focused prevention strategies, and reduce some of the drag that keeps agencies stuck in a reactive cycle. Current work from IACP, NIJ, NIST, Europol, and the Council on Criminal Justice all points in that general direction, even while also stressing the need for careful implementation and safeguards [1][2][3][4][7][11][15][16].
That does not mean every tool is appropriate. It does not mean every vendor claim should be accepted. It does not mean agencies should rush past the concerns. It means law enforcement and its partners should approach AI the same way they should approach any meaningful tool or strategy: with clarity, caution, discipline, training, oversight, and a real commitment to outcomes that matter.
For proactive crime prevention, that means focusing on earlier detection of recurring harm, more precise deployment, better use of existing data and technology, and stronger problem‑solving partnerships—not just adding another system to the stack.
As with any tool or strategy, the question is not simply what AI can do in theory. The question is what can be used in practical, lawful, understandable, and sustainable ways that help agencies and their partners reduce harm, improve public safety, strengthen trust, and make communities—and the work of protecting them—safer.
References
[1] International Association of Chiefs of Police, Artificial Intelligence Resource Hub. https://www.theiacp.org/working-group/artificial-intelligence-resource-hub
[2] National Institute of Justice, Artificial Intelligence topic page. https://nij.ojp.gov/topics/artificial-intelligence
[3] National Institute of Standards and Technology, AI Risk Management Framework. https://www.nist.gov/itl/ai-risk-management-framework
[4] Council on Criminal Justice, National Task Force Releases New Framework to Help Criminal Justice Agencies Assess AI Tools. https://counciloncj.org/national-task-force-releases-new-framework-to-help-criminal-justice-agencies-assess-ai-tools/
[5] National Policing Institute, AI in Policing Beyond the Early Adopters. https://www.policinginstitute.org/infocus/ai-in-policing/
[6] National Institute of Justice, Using Artificial Intelligence to Address Criminal Justice Needs. https://nij.ojp.gov/topics/articles/using-artificial-intelligence-address-criminal-justice-needs
[7] International Association of Chiefs of Police, Implications Associated with Police Use of Artificial Intelligence. https://www.theiacp.org/resources/policy-center-resource/implications-associated-with-police-use-of-artificial-intelligence
[8] National Institute of Standards and Technology, Findings and Recommendations for Field Testing Law Enforcement AI Tools. https://www.nist.gov/document/findings-and-recommendation-field-testing-law-enforcement-ai-tools
[9] National Academies of Sciences, Engineering, and Medicine, Law Enforcement Use of Predictive Policing Approaches: Proceedings of a Workshop. https://nap.nationalacademies.org/catalog/28036/law-enforcement-use-of-predictive-policing-approaches-proceedings-of-a-workshop
[10] National Institute of Justice, Landscape Study of Generative Artificial Intelligence in the Criminal Justice System. https://nij.ojp.gov/library/publications/landscape-study-generative-artificial-intelligence-criminal-justice-system
[11] Europol Innovation Lab, AI Bias in Law Enforcement: A Practical Guide. https://www.europol.europa.eu/cms/sites/default/files/documents/AI_bias_in_law_enforcement_-_practical_guide.pdf
[12] National Institute of Justice, From Crime Mapping to Crime Forecasting: The Evolution of Place-Based Policing. https://nij.ojp.gov/topics/articles/crime-mapping-crime-forecasting-evolution-place-based-policing
[13] National Institute of Justice, Term of the Month: Risk Terrain Modeling. https://nij.ojp.gov/term-month
[14] National Institute of Justice, Police Technologies for Place-Based Crime Prevention: Integrating Risk Terrain Modeling for Actionable Intelligence. https://nij.ojp.gov/library/publications/police-technologies-place-based-crime-prevention-integrating-risk-terrain
[15] Europol Innovation Lab, AI and Policing: The Benefits and Challenges of Artificial Intelligence for Law Enforcement. https://www.europol.europa.eu/cms/sites/default/files/documents/AI-and-policing.pdf
[16] Council on Criminal Justice, Principles for the Use of AI in Criminal Justice. https://counciloncj.org/principles-for-the-use-of-ai-in-criminal-justice/
[17] NIST AI Risk Management Framework resources. https://airc.nist.gov/airmf-resources/airmf/
Comments