Infrared
Transforming digital footprints into actionable evidence for survivor justice.
Background
With nearly 4.8 million victims worldwide, commercial sexual exploitation is a $99 billion underground industry. According to the United Nations, this scourge disproportionately impacts North America. Yet, in Canada, only 10% of cases end in conviction.
The systemic failure towards victims of sexual exploitation stems from many issues. Notably, low conviction rates can be tied to the burden of proof. Currently, the justice system relies too heavily on survivor testimony. Put through a grueling ordeal, discredited over the slightest inconsistency, and subjected to stigma, survivors often emerge from the legal process re-traumatized. This failure creates a chilling effect: it discourages survivors from seeking justice and turns human trafficking into a “low-risk” crime that operates more easily.
The Infrared Tool
Designed for victim services organizations, Infrared supports survivors in seeking justice and rebuilding their lives. This survivor-centered tool equips survivors with the digital footprint of their exploitation to corroborate their accounts, empowering them to press charges, assert their rights in court, secure debt relief, and access healing care. By transforming this data into a powerful asset, Infrared helps survivors of exploitation take back control.
How does Infrared work?
The Infrared process begins with the secure entry of basic information (phone numbers, dates, locations) tied to the persons’ exploitation and the survivor's intake notes. An AI-powered search engine then scans a vast database of automatically archived online advertisements (crawled ads). The algorithm ranks the ads by relevance and automatically groups connected activity together. Finally, an analysis tool cross-references these results with the survivor's account to calculate a "corroboration strength," highlighting matches and discrepancies to establish an picture of the events.
To facilitate the work of caseworkers and ease the burden of proof for survivors, Infrared centralizes case management while ensuring confidentiality, informed consent (strictly tailored to specific needs), and data security through encryption.
Generating evidence reports in the form of secure, temporary links, Infrared is designed to be shared with key stakeholders, such as legal teams. Through this data collection, tools like this provide survivors with the concrete evidence they need to support their court cases and truly access justice.
The Pilot Project With a Community Victim Services Organization
To ensure that Infrared truly meets the realities on the ground, Mila collaborated closely with a community victim services organization. Dedicated to supporting survivors of sexual exploitation, this organization played a key role in the testing and continuous improvement phase of the tool.
During this pilot project, caseworkers tested the platform using mock data (to ensure absolute security and confidentiality during the testing phase). Their feedback helped refine the user interface, optimize risk management, and ensure that the tool seamlessly integrates into the daily workflow of social workers, all while maintaining a strictly trauma-informed approach.
Next Step: Searching for a Host Organization
Today, Infrared is a ready-to-deploy tool. Mila is currently seeking a partner organization to step in as the permanent host and custodian of the tool.
The goal is to transfer the Infrared license to a key player in the social justice sector, such as a large-scale NGO, a legal clinic, or a federation of organizations. This new owner would manage and maintain the platform, use it to support their survivor clientele, and potentially service other victim services organizations across Canada. The ultimate objective is to ensure Infrared’s long-term sustainability, keeping this powerful technology exactly where it belongs: in the hands of those supporting survivors.
Meet the Team
With the collaboration of multiple AI professors and students, as well as legal, ethics and criminology students specialized in the field of human trafficking, sex work and critical perspectives in the domain. The Infrared tool is housed within Mila’s Applied Projects AI for Humanity Studio.
Mila Members
Collaborators
Allison Cohen
Mark Ezema
Frequently Asked Questions
Why use AI in this context, rather than human or traditional methods?
Traffickers post thousands of ads that are quickly deleted. AI makes it possible to analyze a massive database of archived ads, make complex connections, and sort information infinitely faster than a human conducting manual searches.
How can we prevent Infrared from making mistakes with people who are not trafficking survivors?
Infrared is not a surveillance tool that randomly scans the internet to "find" victims. It is only activated once a person self-identifies as a survivor and goes to a community organization to ask for help. The trafficking context is therefore already established. Furthermore, it is the person themselves who reviews the results and validates whether the ads found pertain to them or not, thus preventing false identifications.
How will this data be protected to prevent it from being seen, shared, or used by the wrong people?
In addition to data encryption, Mila has worked with the legal advisors of partner NGOs to develop strict risk mitigation strategies (in the event of misidentification or data leaks) and Privacy Impact Assessments (PIAs). These governance protocols are designed to meet the specific ethical and legal requirements of Ontario and Quebec. Furthermore, the tool provides secure and inclusive authentication methods, even for victims who may not have government-issued identification.
What role should community organizations play in the use or evaluation of Infrared?
Infrared was specifically designed for them. They are the end users of the tool. Moreover, the current goal of the project is to entrust the full management of the tool to a community organization or a legal clinic, which will be responsible for evaluating, maintaining, and offering the tool to other associations.
Resources
T-NET: Weakly Supervised Graph Learning for Combatting Human Trafficking
Proceedings of the 38th AAAI Conference on Artificial Intelligence. This paper propose T-NET, which unlike previous solutions, formulates the problem as weakly supervised classifcation.
Read the report
https://ojs.aaai.org/index.php/AAAI/article/view/30233
Global Estimates of Modern Slavery
The 2017 Global Estimates of Modern Slavery by ILO and Walk Free aim to contribute to the SDGs, particularly Target 8.7, which seeks to end forced labor, modern slavery, human trafficking, and child labor.
Profits and Poverty : The Economic of Forced Labour
This study by the International Labour Organization (ILO) examines the underlying factors of forced labour, one of the main ones being the illegality of profits.