Harnessing Data and Intelligence for Collective Advantage 2026

HPE’s AI ethics assessment of the Global Data Partnership Against Forced Labour Proof of ConceptBOX 7 HPE is committed to ethical, responsible AI. HPE’s AI Ethics Working Group conducted an AI ethics assessment to align the POC with global standards and HPE’s AI Ethics Principles: Privacy-enabled and Secure, Human-focused, Inclusive, Responsible and Robust. HPE’s standard AI ethics assessment is particularly essential to this initiative given the sensitivity of data related to forced labour and its potential wide-ranging, global user base. The assessment follows a structured process, engaging a diverse group of specialists in technology, human rights, risk management and governance. Their role is to provide balanced oversight, challenge assumptions and offer recommendations for embedding ethics throughout the design and implementation. Key recommendations and priority issues for the POC phase: Privacy-enabled: Enforce strict anonymization; avoid personal data; identify and remove personal data; embed privacy-by-design principles. Secure: Employ and encourage other partners to use robust encryption and data protection standards and techniques; establish strong partner organization and user authentication and verification; assign responsibility for auditing data inputs. Data remains in the domain of the data owner, who has exclusive control over usage and access. Human-focused: Restrict user access; facilitate a misuse workshop and adversarial testing before activating the solution beyond the POC; publish responsible use guidance for participants. Inclusive: Be open and inclusive to all regions and participants, their data and use cases; ensure workers or worker representatives trial the pilot and provide feedback on solution- design features; determine an approach for post-POC evaluation of bias risk against vulnerable groups; consider post-POC technical approaches that account for the needs of diverse workers, especially those with weak internet access, mobile-only connectivity or limited English proficiency. Responsible: –Transparency: Build in transparency and explainability for how an AI outcome has been reached, identifying which sources contribute to which outputs. –Accountability: Define responsibilities for all parties through robust governance; maintain continuous ethical oversight; implement logging and accountability measures. Robust: Conduct stress testing and quality assurance upfront and regularly; assign someone to take responsibility for ongoing checks and improvements; collect user feedback to improve tool accuracy. An additional ethics assessment is conducted as the POC transitions into a fully scaled, publicly accessible platform. Harnessing Data and Intelligence for Collective Advantage: Ending Forced Labour in Global Supply Chains 17