This AI Tool is Fighting Fashion’s Uyghur Forced Labor Problem. For Free.
An AI-powered tool designed to sniff—and snuff—out Uyghur forced labor risks in the global apparel supply chain began, in part, as a way to help Laura Murphy catch some Zzzs.
Murphy was until recently a professor of human rights and contemporary slavery at Sheffield Hallam University’s Helena Kennedy Centre for International Justice. When Covid-19 put most of the world in lockdown in 2020, she struck up a working friendship with Shawn Bhimani, an assistant professor teaching supply chain management at Northeastern University, over Zoom.
More from Sourcing Journal
The two talked about worsening U.S.-Sino relations and the deteriorating human rights situation in China’s Xinjiang Uyghur Autonomous Region, where reports of Uyghurs and other Turkic Muslim minorities being thrown into so-called “reeducation” camps, pressed into involuntary servitude and subjected to mental and physical torture had surged. Murphy confessed that she was literally losing sleep over how popular Western brands and retailers were involved in the cultural and religious persecution—even genocide—of vast numbers of people by sourcing products and materials from Xinjiang, however indirectly and unknowingly.
“Laura wasn’t sleeping much at night because she was just investigating facilities night after night after night,” Bhimani recalled. “I got to telling her that machine learning could do that for her so she didn’t have to investigate as much; she could just write about [what she found] and put it in the news. And so that was the inspiration: how can we get Laura to go to sleep?”
Bhimani calls Supply Trace, which was born shortly after the two universities received enough funding from Laudes Foundation to launch a pilot, a “child of the pandemic,” one that was weaned on data points from tens of millions of customs records stretching back a decade. The technology underpinning the database uses a combination of machine learning and human intelligence to connect the dots between buyers and the providers of their goods, creating strings of likely supply chain relationships that highlight potentially problematic nodes. Where applicable, it provides links to supporting evidence such as online media reports or Sheffield Hallam’s prodigious and oft-quoted body of research.
Murphy, who is currently on secondment with the U.S. Department of Homeland Security, wasn’t able to comment due to her new role, but presumably she’s been able to get more shut-eye. What her bleary-eyed conversations with Bhimani have led to is something wholly new, not least because Supply Trace makes information that is usually shut up behind paywalls free and accessible.
“There is a lot of money in this,” Bhimani said. “We’re tracking a couple of dozen platforms that charge between $5,000 and $50,000 a month per user to access this trade data. That means that over 99 percent of the world, including small-to-medium-sized companies, will never be able to afford those monthly rates.”
Not that Bhimani has a problem with for-profit platforms. In fact, Supply Trace hopes to collaborate and trade data with them. The point remains, however, that “the world needs a way to know where our sources are, where this stuff comes from,” he said. For companies scrambling in the wake of the Uyghur Forced Labor Prevention Act, which imposes a rebuttable presumption that all goods made in whole or in part in Xinjiang are the products of forced labor and therefore verboten on the U.S. market, this knowledge has taken on an existential urgency.
“Companies need to be able to do this and the problem is most of them don’t know where to start,” Bhimani said. “And so this allows for companies to have a starting point to begin their due diligence. Now you can see not just your Tier 1 supplier, your Tier 2 supplier on our platform. We’re building out to Tier five in some cases. Companies sometimes don’t have full-time people that know how to do that. They don’t have the time to do that. And they don’t have the resources to do that.”
And not just in the United States, said Lucas A. Rock, an associate involved with import compliance matters at the law firm ArentFox Schiff. As more countries—the European Union, in particular—roll out forced labor import bans similar to the UFLPA, supply chain tracing and risk assessment solutions are becoming increasingly necessary. And though they cannot replace manual supply chain tracing altogether, they can “provide a comprehensive source of data to identify entities within complex supply chains,” he said.
Certainly, there’s always the risk of false positives with AI, especially when you have companies with similar names in similar locales, Bhimani admitted. That’s why “human involvement is so important to be able to scan and check for those,” he said.
It’s for this reason that Supply Trace is deliberate in its use of the term “potential risk,” meaning a connection that’s been flagged by the system but requires attention and verification by a sourcing expert. If it’s made a mistake, the team wants to know, too. The platform’s goal, Bhimani said, is not to “name and shame” companies but rather allow them to do the work they need to comply with the law, not just with apparel and China but potentially with other categories and geographies, as well. There’s no timeline yet, but discussions are beginning to coalesce around what might be possible with the right expert partners.
“It would be great to see the database expand to include data on producers and suppliers using cotton or cotton products originating in Turkmenistan,” said Allison Gill, forced labor program director at the Washington, D.C.-based nonprofit Global Labor Justice-International Labor Rights, a member of a coalition known as the Cotton Campaign to create decent working conditions for cotton workers in Central Asia. “We know many brands are at risk of using Turkmenistan cotton in their products through suppliers in third countries, primarily Turkey and Pakistan.”
Data is a right that everyone should have, Bhimani said. The problem with forced labor monitoring and enforcement today is that everyone is “doing it in silos,” not being transparent or not collaborating, which he finds “unsustainable.”
“’What surprises me the most is that it’s not a given that people can know where their stuff comes from,” he said. “As a prior sourcing expert who worked in global sourcing, I knew where my stuff came from, but my consumers never did. And I find it surprising now that I can see how easy it is to access a wealth of data that has never existed in supply chain history before. But that means it’s just a huge area for opportunity.”
Solve the daily Crossword
