Harnessing AI to fight human trafficking

Human trafficking is the exploitation of people through force, fraud, or coercion to encompass sexual exploitation, domestic servitude, organ removal, and forced labor, marriages, and criminal activity. It is a growing crisis across the globe, with over 27 million victims identified in 2023 (some estimates place this number closer to 50 million worldwide). As this issue grows, so do the difficulties in addressing it, partially due to a lack of consistency in reporting and definitions along with insufficient resources, fragmented coordination, and the constantly evolving and advancing tactics used by traffickers.  

State and local government agencies play a crucial role in combating human trafficking, from major entities like law enforcement, state attorney general offices, task forces, and Child Protective Services, to local community-based organizations. However, many face an uphill battle in developing and implementing effective anti-human trafficking initiatives. One way to offset these restrictions is to enable more Artificial Intelligence (AI) approaches and decision intelligence (DI) platforms to help process data and uncover these situations. 

AI has become a powerful ally for state and local officials and law enforcement agencies grappling with how to address this complex issue. AI-enabled technologies can spot hard-to-detect patterns that often indicate signs of criminal activity related to human trafficking, such as connections to sex trafficking business, watchlists, the use of foreign currencies, money mules, pre-paid cards, shell companies, and extensive use of money service businesses.  

A strong data infrastructure is crucial for officials fighting human trafficking. Integrating diverse data sources—law enforcement, social services, financial records—enables advanced analysis, revealing hidden patterns and complex trafficking networks. Spotting subtle signs, like unusual financial activity, requires this robust foundation. It allows for a proactive, intelligence-driven approach, moving beyond reactive measures to effectively disrupt and dismantle these criminal operations, ultimately protecting vulnerable populations. 

AI-enabled anti-human trafficking initiatives  

Advanced AI tools, fueled by robust, reliable data, are helping uncover connections and relationships between parties involved in human trafficking that have until now, remained hidden. Criminal activity is identified through data patterns such as receiving multiple payments from unrelated people at different locations, frequent large cash deposits, bank accounts opened for multiple customers at the same time that exhibit no normally expected activity (bill payments, payroll deposits, shopping, etc.), spikes in transactions at sites known for high human trafficking risks (hotels, airports, massage parlors, etc), deposits or payments that are intentionally structured under a certain threshold to avoid reporting (ex: $9,900 instead of $10,000), high volume of one-way flight tickets booked by a singular party, or businesses with revolving ownership or address changes indicating potential shell companies. Once exposed, these criminal networks can be targeted and disrupted. 

However, one of the biggest challenges that state and local government agencies and law enforcement face in identifying signs of human trafficking is the sheer volume and diversity of data at their disposal. For example, each year, FinCEN receives over 25 million Bank Secrecy Act transactions (SARs, CTRs, etc) from over 260,000 regulated financial institutions. Money services businesses (Western Union, Money Gram, etc) account for an estimated 500 million annual transactions to all countries except Iran and North Korea. Venmo has over 90 million active users (US) and accounts for over $300B in transfers. The sheer volume of open source, social media, and public record data further compounds the volumes of data to process and analyze. Combined with over 66 million international tourist visitors every year in the United States, the data integration problems become very complex. 

AI has the ability to not only automate timely tasks such as sifting through large amounts of data, but to make connections and spot patterns typically not easily recognizable between financial and non-financial data that may indicate trafficking. However, this requires a foundation of robust, reliable data – an authoritative data source considered the “single source of truth.” AI-powered Entity Resolution provides a critical tool for both de-siloing disparate data and matching duplicate records to create accurate, up-to-date and actionable data points.  

Entity resolution is the process of resolving/unifying two or more data entries (consolidating multiple entities that refer to the same “real world” entity into one). For example, the entities could be a business name, address or individual. This technology links records, de-duplicates and matches data within systems, leveraging both internal, external, structured and unstructured data, improving data quality and accuracy while making it easier to uncover potential relationships between individuals and accounts that may be engaged in criminal activity.  

The final piece of the data puzzle is providing state and local agencies with access to broader content to improve the quality of the data for decision making. This data, overlaid with persona-based typologies and red flag identification protocols, accelerates the identification of potential human trafficking patterns and schemes.  

Both the quantity and quality of data matters when detecting hidden patterns, connections, and relationships that may be potential indicators of human trafficking. Providing access to multiple data sets from various sources maximizes both the perspective and performance of AI models, minimizing data blind spots where criminal interactions may be hidden. 

Why context is critical  

One of the biggest pitfalls for state and local agencies and officials in relying on data to identify human trafficking is lack of context. This context is crucial to make informed decisions. Legacy technology often operates in silos, relying on simple keyword searches, and isolated datasets, which limits its ability to recognize the nuanced patterns that characterize trafficking activity. These limitations mean investigators may only see isolated data points, such as flagged phone numbers, watchlist matches, or certain keywords in online ads, without being able to link them together in a way that reveals a broader criminal network or the shifting tactics of traffickers.  

In contrast, AI can effectively combine and analyze large amounts of data from multiple sources – social media, phone records, hotel bookings, financial transactions – and identify subtle patterns and connections between entities. Knowledge graphs are one example of AI tools used to help define the relationship between entities along with the direction and the strength of the connection. Modeling data in knowledge graphs allows organizations to truly leverage their understanding of relationships between entities to gain important contextual insights, driving intelligent decision-making. 

Context also strengthens AI models’ decision-making ability, leading to higher success rates in accurately identifying instances of human trafficking.  This is also critical for detecting, exposing and uncovering new or emergent trends.  

Tying it all together for a world with less human trafficking  

As the human trafficking crisis continues to grow, it’s a critical issue for state and local agencies to address. Current legacy techniques have major limitations, but the next generation of AI-powered detection technology could be a powerful ally in anti-human trafficking initiatives.  

To advance human trafficking detection techniques and strengthen decision-making through AI, it is critical that state and local agencies have a strong data foundation in place. This foundation makes it possible (with the help of AI) for state and local investigators to parse through large amounts of data and see connections and interactions between entities that indicate potential trafficking, which would likely otherwise go unnoticed. These connections create context, which results in more accurate detection of trafficking patterns. Once this trusted foundation is in place, state and local agencies can use this to fuel powerful AI analysis to reveal new insights into suspicious financial activity, travel events, job-solicitations, social media feeds, or and criminal behaviors to potentially help detect indicators of human trafficking.   

Want to find out more? Get in touch.