A Sound of Hope for South Africa in the Fight Against Trafficking with AI

South Africa faces significant challenges in its fight against trafficking, but technological interventions, especially AI, present a potential turning point in both human trafficking and money laundering prevention.

SwissCognitive Guest Blogger: Gregg Barrett, Chief Executive Officer, Cirrus – “A Sound of Hope for South Africa in the Fight Against Trafficking with AI”

The release of the movie, the Sound of Freedom, while not representative of all the facets of current day human trafficking, has helped initiate broader conversations about modern exploitation. For many who watch the movie the feeling will be that doing something is better than doing nothing. But what can be done that will actually help? I was presented with this question in 2021 by an anti-human trafficking (AHT) NGO operating in South Africa and internationally. Prior, the organisation had been operating at one of South Africa’s busiest airports, and in the course of their work had uncovered systemic corruption and organised crime involved in trafficking at the airport. As a result, their security clearance had been revoked preventing them from operating in the airport. Their cost per intercept was around 300 USD which needed to be reduced to around 50 USD to align with funding. Astutely they believed that turning to the adoption of artificial intelligence (AI) methods would allow them to increase the efficiency and effectiveness of their work, and that this could be scaled without a resulting scaling in cost and headcount. Could there be a greater application of AI than that which results in the saving of lives? [1]

From an institutional governance standpoint, South Africa’s law enforcement institutions are not resourced to the same extent as many of those in the developed western world. [2] Under such conditions trafficking thrives, elevating the role of civil society organisations. However, as with law enforcement, these civil society organisations are not equipped with the optimal tools to enable the application of AI in the fight against trafficking. In understanding the scope and nature of human trafficking in South Africa, the United States Agency for International Development released a first report from a larger authoritative study. Five key points from the report are instructive:

Human trafficking is indeed a serious, pervasive, and systemic problem in South Africa, that seamlessly intersperses with other crimes and social phenomena — including gender-based violence, prostitution, organised crime, missing persons, irregular migration, child abuse and labour disputes, to name a few.
South Africa is not nearly equipped or co-ordinated enough to deal with this crime as effectively as it should or could, and enabling factors such as corruption, complicity, and compromise of officials and other AHT role players is a constant stark background to AHT efforts.
There is poor record keeping, inaccessibility of official trafficking data, and the absence of an integrated information system required to collate and analyse specific information.
A lack of proactive investigations and intelligence sharing, and a largely inactive national AHT task team, means that evidence of South Africa as a transit country is not proactively pursued with international counterparts.
Citing a 2019 multi-country report by the UN, the report says South Africa is a “main destination” for smuggled and trafficked persons on the African continent. According to that 2019 report: “Most Africans see South Africa as the easiest country of transit to reach Europe or the Americas”, and it is “an origin and transit country for trafficking towards Europe and North America, and for trafficking and smuggling to and from Latin America and Asia”.

Returning to 2021 in my engagements with the NGO, the first step became obvious. We needed to focus on the data. Data is the necessary input for the training of AI models and their deployment (inference). In North America for example, Polaris operates the U.S. National Human Trafficking Hotline. Through that work, they have built the largest known dataset on human trafficking in North America, with the data informing real time strategies. Underpinning Polaris is what is known as a data management platform. It is this nontrivial piece of technology that effectively enables the application of AI to AHT.

At a high level a data management platform provides the capability needed to store, manage, share, find and use data for AI. A database is not a data management platform. Rather a data management platform is required for a single point of data ingestion. As the name implies this is to ingest data in all its forms (structured, unstructured, and semi-structured), and with all key data transfer approaches (batch, micro-batch, and streaming). To do this the data management platform provides a highly configurable set of data integration tools that extend far beyond typical extract-transform-load (ETL) or extract-load-transform (ELT) solutions.

In the context of AHT the data management platform provides the operational fabric that enables the access control framework to restrict access to sensitive information at a granular level, ensuring that analysts see only the specific data points that are necessary to complete their work. This ensures that data is being used effectively to have a positive impact, while protecting the privacy of individuals. The platform also enables detailed privacy impact assessments to codify the risks of the data used and the development of mitigation plans.

Beyond this the existence of a data management platform more fully enables the utilisation of mobility data, a financial intelligence unit, a research hub, and philanthropic engineering support. Mobility data is needed to generate heatmaps to identify choke points for placing new stations with human monitors to intercept trafficked persons. The financial intelligence units’ purpose is to train financial services and anti-money laundering staff on the data management platform itself, and allows these professionals to share knowledge, information, and best practices in real-time. In addition, this unit is intended to help survivors of human trafficking get access to banking services that they would not otherwise qualify for because of poor credit and other issues related to their trafficking experience. The research hub is there to bolster the data science / AI capabilities of the AHT organisations by establishing collaborations with academia and industry. This includes a data facility to provide key data to researchers, academics, law enforcement officers and others seeking to deepen knowledge and understanding in the fight against trafficking. Lastly, philanthropic engineering support is intimately connected to a commercial operation, and develops deep, hands-on, and often long-term relationships with NGOs and social sector organisations. This includes on-site engineering to develop foundational understanding of the respective fields enabling the establishment of broader collaborations of players to work on solving social problems. In the context of AHT, philanthropic engineering support provides the capability to drive on-the-ground action with partners that: have the capacity to utilise data management and AI for maximum impact; with organisations that have data and develop data sources; will benefit from AI analysis of the data; and have staff in place to act on the insights to drive action. Importantly, while philanthropic engineering support directly improves and save lives, the engagements also have a powerful impact internally, helping to attract, retain, and engage employees. The challenges being tackled require world class engineers and in South Africa there is a dire need to create opportunities to cultivate this talent – to offer meaningful work and the chance to make a difference.

Yet, here in South Africa a Polaris like data management platform has not been in sight. Enter anti-money laundering (AML). As with human trafficking, South Africa is plagued by money laundering. In February the country was greylisted by global financial crime watchdog the Financial Action Task Force (FATF) for not fully complying with international standards around the prevention of money laundering, terrorist financing and proliferation financing. Presently, each bank in South Africa undertakes AML mostly in isolation. A recent Bank for International Settlements project confirmed that collaborative analysis and learning approaches were more effective in detecting money laundering networks than the current siloed approach in which financial institutions carry out analysis in isolation. By the numbers, a LexisNexis report found South Africa’s largest banks spent on average 12.3 million USD in financial crime compliance operations in 2021. Further, the total projected cost of financial crime compliance in South Africa increased by 65% between 2019 and 2021, from $2.3 billion USD to $3.8 billion USD.

Contrasting deficiencies in human trafficking and money laundering show significant overlaps. This is unsurprising given that both fundamentally involve intelligence gathering, requiring the very same technological toolsets. Unquestionably, AI is the future for AML and AHT, and South Africa requires an industry wide approach – an industry wide data management platform. Just such an intervention has been initiated, where the intention is that the data management platform and its related operations for AML will be placed into a shared entity co-owned by industry. There is precedent for such. The country’s banks collaborate and co-own Bankserv, the automated clearing house. For South Africa’s largest banks an industry wide AML platform will result in significantly enhanced AML capabilities and cost reductions to around one third of their current financial crime compliance spend. For the country, eventual removal from the FATF greylisting. And for South Africa’s AHT organisations, the provision of a world class platform enabling the application of AI to fight trafficking.

In summary, out of South Africa’s struggles arises an opportunity for the country to be a global leader in the fight against trafficking and money laundering. Technological interventions like this do not come to pass by chance but through leadership. For what is it if we have these technological capabilities but fail to implement? Do we not have an ethical obligation? We must be pragmatic and assume that based on experience Government leadership is unlikely to materialise, and be optimistic that there are leaders in industry with the vision and fortitude to pursue this endeavour that will ultimately save lives. That there are those who are motivated in their hearts and minds to actually do something that will make a difference.

[1]

Ways in which long-range research in AI could be applied to the fight against human trafficking, see:

Bliss, N. et al. (2021). CCC/Code 8.7: Applying AI in the Fight Against Modern Slavery. https://arxiv.org/abs/2106.13186

An algorithm to identify similarities across escort ads, making it easier for law enforcement to identify human traffickers, see: Carnegie Mellon University. (2021). Algorithm Uses Online Ads To Identify Human Traffickers. https://www.ml.cmu.edu/news/news-archive/2021-2025/2021/april/machine-learning-ai-algorithm-uses-online-ads-identify-human-traffickers.html

Algorithms to extract signatures in images, such as specific tattoo designs linked to human trafficking networks, see: MIT News. (2021). Turning technology against human traffickers. https://news.mit.edu/2021/turning-technology-against-human-traffickers-0506

Using AI to reveal trends in payments and help identify victims of modern slavery, see: World Economic Forum. (2020). How AI can help combat slavery and free 40 million victims. https://www.weforum.org/agenda/2020/11/how-ai-can-help-combat-modern-slavery/

Examples of the work done by the Stanford Human Trafficking Data Lab to combat human trafficking, see: Stanford University. (2021). Melding Artificial Intelligence and Algorithms with Health Care and Policy to Combat Human Trafficking. https://fsi.stanford.edu/news/melding-ai-and-algorithms-health-care-and-policy-combat-human-trafficking

[2]

For an index of governance in Africa see: Mo Ibrahim Foundation. (2020). 2020 Ibrahim Index of African Governance: Key Findings. https://mo.ibrahim.foundation/news/2020/2020-ibrahim-index-african-governance-key-findings

For a global comparison across various governance indicators see: The Worldwide Governance Indicators project. (2020). Worldwide Governance Indicators. https://info.worldbank.org/governance/wgi/

With an average score of 32, Sub-Saharan Africa is the lowest performing region on the Transparency International Corruption Perceptions Index, see: Transparency International. (2021). CPI 2020: Sub-Saharan Africa. https://www.transparency.org/en/news/cpi-2020-sub-saharan-africa

About the Author:

Gregg Barrett is a seasoned executive with extensive and diverse experience in strategy, building and managing relationships, deal-making, communication, developing high-performance teams, organisational leadership, and problem-solving across a range of areas. Over the last decade, Gregg has led work in data science, machine learning, corporate research, and corporate venture capital. This includes the establishment and management of data science, machine learning, corporate research, and corporate venture capital operations, working across people, processes, and technology, integrating structured and unstructured data to direct research, business, and investment strategy.