Machine learning is pulling back the curtain on sex trafficking — revealing how fake job ads are used to lure victims into exploitation.
The UN describes human trafficking as: “the recruitment, transportation, transfer, harbouring or receipt of people by force, fraud or deception to exploit them for profit.”
It says the extent of the crime is difficult to ascertain. While about 50,000 cases were reported to the United Nations Office on Drugs and Crime (UNODC) in 2020 by 141 countries, the UN Labor agency (ILO) says 50 million people were living in “modern slavery” in 2021, of these 28 million were in forced labour and 22 million were trapped in forced marriage.
The UN says human trafficking is one of the fastest-growing crimes, along with drugs and arms trafficking, and a highly profitable business, generating an estimated $150 billion in profits each year.
Ending this scourge is complex, with the ILO saying options include improving and enforcing laws and labour inspections; ending state-imposed forced labour; stronger measures to combat forced labour and trafficking in business and supply chains; extending social protection, and strengthening legal protections, including raising the legal age of marriage to 18 without exception.
Now, a new study from the University of Pennsylvania has used machine learning to unmask the deceptive recruitment tactics at the heart of global sex trafficking networks. By analysing millions of online ads, the researchers have mapped the covert pathways traffickers use to move victims from one location to another — and how these routes can now be disrupted before exploitation occurs.
“The AI system scans millions of online posts advertising commercial sex to spot deceptive recruitment tactics, like when the same entity posts a seemingly innocent job offer, like modelling or massage in one location but simultaneously advertises sex sales somewhere else”, says Assistant Professor Hamsa Bastani from the University of Pennsylvania. “By connecting these deceptive posts across locations, the AI helps authorities uncover hidden pathways of potential sex trafficking.”
The study finds that victims are often recruited from suburban and rural areas, far from where the sex sale occurs. That disconnect, the researchers argue, has allowed trafficking to flourish largely unnoticed in smaller, struggling communities.
“By combining data science with deep web analysis, we are helping to uncover trafficking networks and provide law enforcement with tools to intervene before exploitation occurs,” says Bastani. “Our research reveals the hidden supply chains of sex trafficking, showing how recruitment often begins with false promises in vulnerable communities.”
indicate more posts.
The project harnessed real-world data from the deep web and was years in the making. The AI-driven system doesn’t just support law enforcement; it also offers a framework for social service providers, community organisations, and policymakers to work together in identifying and protecting at-risk individuals.
“What makes this research so powerful is its scalability and real-world application,” says Bastani. “We’re providing a way for law enforcement to detect trafficking earlier in the process, potentially saving countless lives.”
As trafficking continues to pose a persistent and global challenge, the team hopes their work will pave the way for smarter, more proactive policies. By tracing the recruitment trail back to its origin, their findings offer a promising step toward shutting down exploitation before it begins.
These findings were published in Manufacturing & Service Operations Management.
AI and violence
