Artificial intelligence (AI) has emerged as a transformative tool, offering enhanced capabilities to detect, investigate and disrupt trafficking-related financial crimes. Criminals are just as effectively, and in many cases more adeptly, leveraging these advanced technologies to recruit victims, streamline operations and obscure illicit proceeds from human trafficking (HT). It is imperative to stay ahead of these rapidly evolving criminal methodologies, lest the progress being made will become fruitless.

Artificial intelligence (AI) has emerged as a transformative tool, offering enhanced capabilities to detect, investigate and disrupt trafficking-related financial crimes.¹ Criminals are just as effectively, and in many cases more adeptly, leveraging these advanced technologies to recruit victims, streamline operations and obscure illicit proceeds from human trafficking (HT). It is imperative to stay ahead of these rapidly evolving criminal methodologies, lest the progress being made will become fruitless.

Organizations worldwide are spearheading promising AI research programs and initiatives to tackle HT across multiple disciplines. Examples of AI applications are shown in Graphic 1 below.

Graphic 1: Examples of AI Applications for the Fight Against HT

Source and visualization by: RedCompass Labs

The challenge lies in ensuring these organizations work in sync to maximize the potential of these innovative methods and tools. This article will examine investigative enhancements, typologies enhancements, contextual monitoring and detection, and nontraditional intelligence integration more closely.

Investigative Enhancements

AI has transformed the investigative process by significantly accelerating the analysis of financial transactions. AI-powered tools can sift through vast amounts of transactional data in real time, identifying subtle and complex patterns that traditional systems often overlook. This capability enables investigators to quickly connect transactions across multiple accounts, financial institutions (FIs), and jurisdictions, facilitating a more comprehensive analysis and expediting the detection of HT activities. Examples include:

• Account takeovers are a method traffickers frequently use to exploit victims’ financial accounts. Further exploitation can be prevented by identifying anomalies such as unusual login behaviors, sudden changes in transaction activity or deviations in withdrawal patterns.
• Training programs incorporating AI-powered insights, simulations and real-world case studies are providing investigators with immersive, hands-on learning experiences, leading to greater recognition and faster, more effective responses to trafficking indicators.
• Natural language processing (NLP) is being used at scale to focus on adverse media and unstructured news articles, legal documents and social media to identify individuals and/or entities with potential links to trafficking activities.
• Predictive analytics are being used to analyze historical transaction data to proactively identify patterns that point to high-risk activities or locations. FIs and law enforcement (LE) have been able to anticipate trafficking risks and intervene early, safeguarding potential victims and preventing further exploitation.

Nontraditional Intelligence

Nontraditional intelligence has been defined as the use of information sources that are not typically used for the specified purpose. Traditional intelligence (e.g., negative news, corporate records) is readily available from large-scale pay-to-play data providers. Nontraditional intelligence provides more defined insights that allow for a more effective approach to understanding the problem set more completely. Therefore, it allows for a more effective approach to detecting, deterring and disrupting the activity from occurring in the first place. However, while nontraditional intelligence is readily available, it is harder to source and not typically employed among traditional intelligence sources or technology to counter HT.

Why is the use of nontraditional intelligence so important in the cause of battling HT? HT is a broad issue covering many geographic areas and involving many existing threat groups. HT also employs a variety of techniques in the execution of the crime, so it requires a diversity of intelligence in order to completely understand the issue as well as the who, what, when, where, how and why. The answers to these questions are essential in responding to this complex global threat.

Nontraditional intelligence encompasses public domain data from various sources such as websites, blogs, social media and the dark web. This data can include personal and business information like names, aliases, contact details and addresses. The subjects of this data may be potential victims or sex workers, both voluntary and involuntary, as well as individuals involved in HT. Leveraging generative AI (GenAI) and large language models (LLMs)² to analyze this diverse and rich data provides a comprehensive understanding of the issue, enabling more complete and informed responses to prompts.

Persona-Based Typologies Enhancements

Maintaining up-to-date HT crime typology libraries or knowledge repositories is a well-known challenge for investigative operations. These libraries, which cover a spectrum of HT crimes, including sex trafficking, labor trafficking, child sexual exploitation, forced criminality, trafficking for organ harvesting and other crimes, require regular updates. The process of identifying emerging trends, extracting and interpreting insights, and deduplicating and implementing new risk indicators can be labor-intensive and inefficient.

AI provides a significant advantage by automating these tasks, ensuring that even the most complex typology libraries remain current and actionable. Additionally, large language models (LLMs) simplify reporting by analyzing large datasets, extracting key insights and generating comprehensive reports, thereby helping teams identify patterns, root causes and thematic issues across different areas.

GenAI, in particular, has the potential to dramatically reduce manual effort in creating and updating structured HT typologies. Through the use of LLMs and other tools (e.g., Microsoft Azure), vector embeddings³ are being generated to represent HT typologies, personas and red flags. AI-powered semantic NLP analysis enables the identification of overlapping patterns, context and similar behaviors, streamlining processes and significantly reducing manual work. Furthermore, LLMs can map the networked relationships between red flags and HT personas, identifying gaps in knowledge and documentation and enhancing overall precision.

AI holds significant potential in enhancing HT intelligence collection. Integrating cutting-edge research, like agentic AI⁴, can lead to solutions capturing new data patterns and behaviors, potentially reducing the need for constant human intervention. While still largely conceptual, curiosity-driven AI mechanisms⁵ could one day autonomously explore data for unknown patterns that may otherwise go unnoticed. Using comprehensive data sources—including transactional data, know your customer information, industry reports and open-source intelligence—AI solutions will broaden HT identification strategies. Enhancing the persona-based typologies framework with AI will lead to more autonomous and adaptive systems.

Contextual Monitoring

Contextual monitoring, underpinned by AI and machine learning, is revolutionising the way FIs generate actionable intelligence. The ability to join and connect data from different systems and sources (internal and external) to create context and meaning is enabling investigators to automatically analyze behavioral anomalies, geographic inconsistencies, historical behavior profiles and more. Ultimately, this allows for the prioritization of genuine threats, improvement in the efficiency and effectiveness of investigative operations, and the reduction of false positives.

Two key technologies are driving contextual monitoring:

• Entity resolution (ER): AI-driven ER is the consolidation of disparate internal (customer and transaction data) and external (traditional and nontraditional) data points, including those tied to HT activities, into unified profiles. This capability enables the identification of individuals using multiple identities, the linking of accounts to trafficking networks and the detection of patterns indicative of exploitation. For instance, AI can unify fragmented data on accounts operated by traffickers, revealing the full scope of their activities.
• Network analysis: AI-driven network analysis maps relationships within complex transactional and nontransactional ecosystems. It illuminates the flow of funds, connections between traffickers, victims and facilitators, and links to high-risk jurisdictions. By visualising these networks, institutions can target key actors and disrupt broader operations, moving beyond a transaction-centric approach to address entire criminal enterprises.

The adaptability of AI and machine learning ensures that detection systems remain relevant in the face of evolving criminal tactics. These technologies continually refine their models by learning from past investigation and incorporating new patterns of suspicious behavior to ensure frameworks remain robust. Many organizations are further strengthening contextual monitoring by incorporating persona-based monitoring scenarios and nontraditional intelligence. See Graphic 2 below for an example of a real-life case study where persona-based monitoring scenarios and nontraditional intelligence automatically identified a sex trafficking network that the previous technology did not uncover.

Graphic 2: Example of a Real-Life Case Study Involving Contextual Monitoring

Source and visualization by: Chris Bagnall

Cross-Sectoral Collaboration

The future of AI in combating HT extends beyond individual institutions, offering a pathway to bridge gaps and more effectively collaborate across sectors creating a unified front against HT.

Secure AI-driven platforms such as federated learning models are enabling FIs, LE agencies and nongovernmental organizations to share intelligence while maintaining compliance with data privacy regulations. By aggregating insights from diverse sources, these platforms provide a more complete picture of trafficking networks, uncovering connections that might otherwise remain hidden.

For example, AI is correlating data from banking systems, LE records and victim support organizations to map trafficking networks across multiple jurisdictions. Such collaboration not only accelerates investigations but also ensures that interventions are informed by a holistic understanding of the trafficking ecosystem. Additionally, shared AI tools are facilitating joint training initiatives, allowing stakeholders to build collective expertise and align strategies.

One such example is the Nebraska Human Trafficking Task Force (NHTTF)⁶ founded in 2015. The NHTTF coordinates the state’s response to HT occurring in Nebraska in collaboration with LE and prosecutors, but also FIs, service providers, advocates and community partners. The NHTTF has been recognized as a leading model and approach to tackling HT at a state level. Many states in the U.S. have worked with the NHTTF to adopt or improve their own task force models.

HT AI Research Projects

For further insights, there are several prominent academic research programs dedicated to applying AI to combat HT. Notable initiatives include:

1. The Alan Turing Institute⁷
   The U.K.’s national institute for data science and AI has conducted research on leveraging AI to combat modern slavery and HT. Their projects explore how machine learning can detect trafficking patterns within large datasets, enhancing early detection and intervention strategies.

2. The Southern Methodist University (SMU) Human Trafficking Data Research Project⁸
   This SMU endeavor serves as a central hub for securely storing, enriching and utilizing HT data. It facilitates access to overlapping datasets, enhancing data efficacy, and provides secure access to LE, policymakers, researchers and other authorized stakeholders.

3. The Center for Research and Education in Counter Human Trafficking (CRECHT) at the University of Houston⁹
   CRECHT aims to leverage interdisciplinary expertise, advanced technologies and innovative methodologies to combat HT. This center is dedicated to developing research-driven strategies that support prevention, detection and survivor recovery while also fostering partnerships with various sectors, including LE, FIs and policymakers.

4. The Modern-Day Exploitation Lab at QSIDE Institute¹⁰
   This lab integrates advanced analytical methods, including AI and data science, to understand and address modern-day exploitation. By analyzing complex datasets, the lab seeks to uncover patterns and networks associated with HT, informing policies and intervention strategies.

Real-World Application

The transformative potential of AI in addressing HT is best illustrated through real-world applications. By leveraging AI-driven tools, FIs and LE agencies have made significant strides in identifying and disrupting trafficking networks, as demonstrated in the following case studies.

Case Study 1: Disrupting a Trafficking Network

An FI employing contextual monitoring detected a series of small, structured deposits across multiple accounts. Upon further investigation, enhanced entity resolution revealed that these accounts were connected to a network of recruiters and transporters operating across multiple countries. Using advanced network analysis, the institution uncovered links to businesses in high-risk industries, such as massage parlors and nail salons, often associated with HT operations.

By sharing these insights with LE, the FI facilitated the dismantling of the trafficking network. The coordinated effort led to the rescue of dozens of victims and the arrest of key perpetrators, demonstrating the power of AI in identifying and addressing trafficking activities at scale.

Case Study 2: Identifying Victim Exploitation

An AML analyst flagged an account with frequent international wire transfers to jurisdictions known for HT. Contextual monitoring raised red flags as the account holder was a low-income individual with no clear justification for such transactions. Enhanced network analysis linked the account to a broader network of traffickers exploiting victims for forced labor.

This insight spurred a deeper investigation, uncovering the scale of the operation and enabling authorities to intervene. The resulting collaboration between the FI and LE led to the identification and prosecution of the traffickers, as well as the recovery and support of victims.

Case Study 3: Understanding HT Typologies

The AI assistant in the HT typologies portal (knowledge repository) functions as an intelligent support system helping over 80 FIs worldwide to enhance their effectiveness in identifying and combating HT and related financial crimes. It is custom-built and operates as a conversational and research tool, leveraging advanced GenAI capabilities to interact with users and provide tailored responses, guidance and resources.

Key functionalities include contextual search, interactive knowledge access, guided investigations and training. The AI assistant's integration into the portal ensured that FIs leverage advanced technologies to strengthen their defenses against trafficking and enhance operational efficiency and compliance.

Conclusion

Advanced technologies, particularly AI, are revolutionizing efforts to detect and disrupt HT. These tools enable earlier interventions, reduce the societal impact of HT, and disrupt criminal operations at scale.

However, challenges remain. Implementing AI-driven solutions requires high-quality data, robust governance frameworks and careful navigation of privacy regulations. Organizations must strategically plan, collaborate across sectors and integrate nontraditional HT risk data to maximize AI’s potential.

The fight against HT is a collective effort and AI plays a crucial role in fostering a safer, more transparent ecosystem. By balancing innovation with ethical considerations and human oversight, stakeholders can strengthen defenses against exploitation. As technology advances, so does our ability to protect vulnerable populations and hold perpetrators accountable. Through continuous enhancement and collaboration, we can drive meaningful change and offer hope to millions of victims worldwide.

Chris Bagnall, CAMS-FCI, CFE, head of mid-size banking solutions, Quantexa,

Silvija Krupena, CAMS, CFCS, director, financial intelligence, RedCompass Labs, RedFlag Accelerator,

Chris Kemp, director of enterprise operations, Anti-Human Trafficking Intelligence Initiative,

Freddy Massimi III, senior enhanced due diligence associate, Guidehouse,

1. “New frontiers: The use of generative artificial intelligence to facilitate trafficking in persons,” Organization for Security and Co-operation in Europe, November 1, 2024, https://www.osce.org/cthb/579715
2. “What are large language models (LLMs)?” Microsoft, https://azure.microsoft.com/en-us/resources/cloud-computing-dictionary/what-are-large-language-models-llms
3. “Vector Embeddings in RAG Applications,” Weights & Biases, https://wandb.ai/mostafaibrahim17/ml-articles/reports/Vector-Embeddings-in-RAG-Applications--Vmlldzo3OTk1NDA5
4. “Agentic AI: 4 reasons why it’s the next big thing in AI research,” IBM, October 11, 2024, https://www.ibm.com/think/
   insights/agentic-ai
5. Ben Lutkevich, “Curiosity artificial intelligence (curiosity AI),” TechTarget, https://www.techtarget.com/whatis/
   definition/curious-AI
6. “Nebraska Human Trafficking Task Force,” Mike Hilgers, Nebraska Attorney General, https://ago.nebraska.gov/nebraska-human-trafficking-task-force
7. “Data science for tackling modern slavery,” The Alan Turing Institute, https://www.turing.ac.uk/research/research-projects/data-science-tackling-modern-slavery
8. “About the Human Trafficking Data Research Project,” SMU Dedman College of Humanities and Sciences, https://www.smu.edu/dedman/research/htdr/about
9. “The Center for Research and Education in Counter Human Trafficking (CRECHT) homepage,” University of Houston, https://crecht.egr.uh.edu/
10. “The Modern-day Exploitation Lab homepage,” QSIDE Institute, https://qsideinstitute.org/modern-day-exploitation-lab/