How Amazon is Using AI to Identify Human Rights Risks

Amazon has begun deploying AI models capable of identifying forced labour risks across its vast supplier network.

The company's machine learning system analyses millions of data points from historical audits, government reports and news signals to flag high-risk supplier sites.

Kara Hurst, Amazon's Chief Sustainability Officer, says that the tool "successfully identified about 9 out of every 10 high-risk sites, with 85% overall accuracy".

This allows the company to prioritise due diligence resources more effectively across its complex operations, which span e-commerce, logistics, cloud services and manufacturing.

Amazon has also developed an AI tool that processes audit reports in minutes rather than the four hours typically required for manual review.

Kara said early versions helped process audit reports "65% faster - a remarkable difference".

From principles to practice

Amazon formalised its enterprise-wide human rights commitment in 2019 with global human rights principles aligned to UN guiding principles on business and human rights. The company had published its first supplier code of conduct five years earlier in 2014.

Leigh Anne DeWine, Amazon's Director of Human Rights & Social Impact, told Devex that real progress has come from operationalising those principles through collaboration with individual business units.

She said the company moved from enterprise-wide human rights assessments to building tailored due diligence processes that reflect unique risk profiles across different sectors.

The retailer has also expanded supplier transparency by publicly mapping suppliers on Open Supply Hub.

Last year Amazon addressed 100% of the 826 complaints received through its human rights and environmental complaints form.

The challenge of data fragmentation

Leigh Anne sees data fragmentation as the most persistent obstacle facing businesses attempting to advance human rights protections.

She explained that human rights issues often remain hidden because information is "incomplete, inconsistent, or siloed across owners, countries and systems".

Without shared data standards and trusted collaboration mechanisms, efforts to address risk remain siloed and reactive.

Companies may identify issues in one facility or region but lack the interoperability to act collectively or upstream where many risks originate.

This challenge prompted Amazon to join the World Economic Forum's Global Data Partnership Against Forced Labour, which aims to accelerate responsible data sharing across sectors.

Coalitions in support of human rights

Amazon co-founded Tech Against Trafficking, a coalition that includes Google, Microsoft, Meta and TikTok.

The initiative unites industry, civil society, technologists and survivors to scale technology solutions combating human trafficking.

Amazon has contributed technical expertise and AWS cloud solutions to support anti-trafficking NGOs.

The company hosted TAT summits in 2022 and 2024 and helped advance a benchmark on how companies collect forced labour indicators.

Another partnership with the International Organization for Migration focuses on strengthening ethical recruitment practices in countries of origin and destination.

This addresses worker-paid recruitment fees, deception and wage theft that can indicate forced labour and often begin during recruitment and migration processes.

The limits of AI

Despite the promise of AI-powered risk detection, Leigh Anne emphasised that technology must complement rather than replace human judgement.

She believes that "even the most advanced models are only as strong as the data behind them".

Two challenges stand out: expanding the breadth and quality of training data, and ensuring the broader industry adopts AI responsibly.

Amazon collaborates with NGOs, research organisations and audit firms to improve data coverage and rigorously test its models.

Amazon's strategic priorities when it comes to human rights

The company's human rights priorities centre on several pillars: strengthening policy foundations, deepening business integration into sectors like logistics and construction, advancing AI-powered risk insights, expanding stakeholder engagement and strengthening grievance mechanisms.

Workers' voices remain central to this approach.

As Amazon works to decarbonise operations, it aims to ensure the transition is just by integrating social considerations into environmental strategies.

Leigh Anne defined long-term success as embedding human rights in every decision and creating conditions where forced labour becomes increasingly rare.

She said progress will be evident when "the most vulnerable workers in global supply chains experience greater agency, safety and remedy".