By cryptoadmin 
In 2017, Kubiiki Pride’s desperate search of 270 days to find her 13-year-old daughter led her to the world of the dark web. She rummaged through hundreds of advertisements on backpage.com that hosted around 70% of online sex ads in the US market until she found her daughter featured as an escort available to be rented out, where she was disturbingly described as ‘young and new‘.
The link with stars and hearts struck her, which she then clicked to find explicit photos of her daughter. The mother had to buy the service to get her daughter back, who was by then addicted to drugs and had been brutally abused.
Human trafficking encompasses recruitment, transportation, entrapment, brokering, delivery, and exploitation of victims. Law enforcement and government agencies have been using artificial intelligence to search through labyrinths of data points to fight this heinous crime.
One of the major tasks for authorities is identifying clues of human trafficking in online sex adverts as thousands of them are posted every week. However, the distinction between human trafficking risk and consensual sex work can be identified through the use of AI and machine learning algorithms by studying the deceptive recruitment behaviours linked to sex sales visible in deep web data.
Law enforcement has increasingly monitored websites like skipthegames.com, backpage.com and others, with 14 million records analysed for suspicious activity.
These records contain post text, location details, phone numbers, and metadata, revealing recruitment strategies leading to sex sales. For instance, recruitment ads for escort or modelling services, devoid of explicit sex service mentions, are linked to the same contact information in multiple sex sales ads, signalling potential deception and trafficking risks.
“We’ve learned that each organisation can have multiple templates they use when they post their ads, and each template is more or less unique to the organisation. By template matching, we essentially have an organisation-discovery algorithm,” says Dr. Lin Li from the Artificial Intelligence (AI) Technology Group at MIT Lincoln Laboratory.
Collaborations between academia and law enforcement agencies are advancing this aspect. The Lincoln Laboratory, led by DR. Lin has developed algorithms to extract specific signatures from images associated with trafficking networks. It involves leveraging machine learning to identify potential trafficking activities within online sex ads, even in decentralised platforms. They’ve also created systems to analyse digital evidence, from text to imagery and audio, making it easier for investigators to connect the dots and build cases efficiently.
Additionally, Researchers at Carnegie Mellon University have developed an AI-based tool called Traffic Jam, which compares images uploaded by hotel guests to a database of known trafficking locations. It uses facial recognition and geospatial software to identify missing individuals in online ads.
DARPA’s Memex program and IBM’s Traffik Analysis Hub also analyse vast datasets from social media, online ads, and the dark web to detect high-risk locations and patterns.
Identifying Supply Chains
Hamsa Bastani, a Professor at the Wharton School, recently presented a summary of her ongoing work using machine learning and Snorkel AI’s tools, which helps detect and track activities that are associated with a high risk for global sex trafficking—including the analysis of recruitment-to-sex-sales pathways, offering insights into the complex networks facilitating human trafficking.
High-density edges in the visualisation signify significant movement between locations, showcasing the intricate flow of trafficking activities. Expected patterns, such as victims recruited in the Midwest being sold on different coasts in the US and the trafficking of Eastern European women, align with prior anticipations and with mentions of supply chains in India
The goal is to label new posts and discover unique recruitment patterns to understand trafficking networks comprehensively.
This three-part methodology involves domain-informed vocabulary creation, word embedding, and weak learning to form an ensemble model using Snorkel. Expert labelling forms a balanced dataset, enabling the active learning pipeline to identify new recruitment patterns beyond previous knowledge. Finally, metadata linking recruitment to sales posts helps unravel the trafficking supply chain, which is vital for policy implications and law enforcement strategies.
This analysis is also pivotal for law enforcement agencies, offering a strategic approach to combating trafficking. It recommends coordination between jurisdictions involved in recruitment (point A) and sales (point B) to disrupt the supply chain effectively. By targeting the problem from both ends, law enforcement can enhance the success rate of arrests without overburdening resources. This method of targeted collaboration aligns with the observed trafficking flow and helps direct law enforcement efforts more efficiently.
AI-powered platforms like the Global Emancipation Network’s Minerva are already facilitating cross-border collaboration among anti-trafficking organisations, streamlining communication and data sharing.
Drying the Money Flow
Identifying money trails through Bitcoin pathways or banks is another important aspect in this fight, which helps thwart the financial backing of such endeavours, resulting in their collapsing completely.
After being trafficked from Hungary to Canada at the age of 21, Tamir, one such survivor turned advocate, recognised the importance of these transactions in exposing the workings of human traffickers’ operations. She implored financial institutions to join the fight, which prompted Peter Warrick, the director of anti-money laundering risk intelligence at the Bank of Montreal, to rally Canada’s top five banks and the financial regulator to launch “Project Protect.”
The evolving landscape of financial technology, or FinTech, has introduced a new frontier in combating crime. Companies like QuantaVers are at the forefront of using machine learning technology to detect potential criminal activities within financial transactions.
Organisations like Liberty Asia also bridge the gap by providing critical on-the-ground intelligence to banks, enabling them to track and identify traffickers globally because such operations extend beyond borders.
This crackdown on illicit funds laundered through the financial system has resulted in fines exceeding $321 billion over nine years for the banks in an effort to push them towards bolstering their anti-money laundering efforts. However, the success in uncovering trafficking networks represents only a fraction of the success in undermining this elicit industry, which is worth a whopping $150 billion every year.
The post Now AI Can Help Fight Human Trafficking appeared first on Analytics India Magazine.