How tech giants are protecting victims' privacy in the fight against human trafficking

How can we measure a crime as pernicious and brutal as human trafficking without endangering the victims behind the numbers?

Stacker compiled resources from the United Nations International Organization for Migration, or IOM, to explain how new discoveries in statistical analysis are helping to shed light on the dark world of human trafficking.

The UN first set out to resolve this dilemma by partnering with major tech companies, including Microsoft, Amazon, British Telecommunications and Salesforce, through its Tech Against Trafficking Accelerator program in 2019. It released its first anonymized dataset in 2021 and updated it again with case data from 2022.

For decades, widely shared statistics on human trafficking have been limited to estimates based on information reported to organizations such as the National Center for Missing and Exploited Children, or criminal justice statistics, which , experts say, paint a conservative picture, highlighting only cases in which authorities are misinformed. capable of intervening and prosecuting the guilty.

And the available data suffers from another problem: Sharing detailed information about the demographics of people who participate in and are victims of crime risks putting victims in greater danger.

In the 2022 update, the UN's partnership with Microsoft and others launched a new way to shed light on the relationships between victims and their captors while preserving victims' privacy and integrity underlying data. They achieved this by combining cutting-edge statistical methods such as synthetic data and differential privacy.

“Safety in noise”

Synthetic data is data that is artificially generated via algorithms and comes in various forms. It can be based on real data, partially based on real data, or completely artificial. Financial companies like American Express, as well as those in the autonomous vehicle and healthcare industries, have been working to harness the power of synthetic data for several years.

Differential privacy is a method of describing patterns within a group of people without disclosing identifiable information about any individual in the group. Microsoft has used differential privacy algorithms since at least 2017 to protect user privacy when collecting data on their devices.

In the past, agencies releasing case data could remove aspects that could be used to identify an individual. This is also called anonymized data. However, each deletion or modification removes information that could contribute to the understanding of traffic networks.

Other methods of masking data to ensure privacy include aggregated data and anonymized data, a method that studies have shown sometimes distorts the underlying data.

The tech coalition's now-public algorithm ingests sensitive data about U.N. affairs and produces numbers that cannot be linked to the individuals represented there. The authors of the dataset call this “safety in noise.” It is important to note, however, that the data generated by the updated algorithm preserves statistical relationships between cases so that experts can trace patterns without sacrificing victim safety.

The dataset released as part of the accelerator includes records of 17,000 victims as well as more than 37,000 human trafficking perpetrators who engaged in criminal activity in 123 countries from 2005 to 2022, according to IOM.

Specifically, the dataset allows more people to analyze the relationships between trafficking victims and their perpetrators.

Analyze relationships

Synthetic data created based on descriptions of trafficking victims assisted by IOM reveals that forced labor was seven times more prevalent than sexual exploitation from 2021 to 2022, the most recent period for which data is available .

Women were the most common victims of the crime last year, and those victims' traffickers were almost equally likely to be men or women. The perpetrators in these cases were most often strangers to the victim, but a significant portion of victims reported knowing the perpetrator either as a friend or acquaintance.

Human trafficking is a global problem and often occurs when vulnerable populations like migrants are forced into slavery or sex trafficking. Case data that describes the situation of each victim or perpetrator, such as that held by immigration agencies around the world, can potentially help authorities gain information as they work to catch and prosecute perpetrators and to protect victims.

The Accelerator's efforts in collaboration with the UN are part of the Biden administration's updated four-pronged action plan to combat human trafficking, released in 2021. The plan highlights the importance of partnerships with the private sector to increase information sharing that supports efforts to prevent human trafficking. crime, protect victims and prosecute those who facilitate crime.

Story editing by Ashleigh Graf. Copy edited by Paris Close. Selection of photos by Clarese Moller.