Legal & Ethical Issues of AI in Business

Legal and Ethical Issues of the AI in Business Author’s Name The Institutional Affiliation Course Number and Name Instructor Name Assignment Due Date Table of Contents TOC \o "1-3" \h \z \u Summary of the Problem Identified PAGEREF _Toc205325092 \h 3Legal and Ethical Issue Analysis PAGEREF _Toc205325093 \h 3Descriptive Examples of the Issue PAGEREF _Toc205325094 \h 4Business-Risk Assessment PAGEREF _Toc205325095 \h 5Stakeholders Harmed PAGEREF _Toc205325096 \h 6Prevention and Mitigation Recommendations PAGEREF _Toc205325097 \h 7References PAGEREF _Toc205325098 \h 8 This report critically evaluates the uses of AI-enabled supply chain tools, such as those made by Altana and FRDM.ai, in identifying the human rights risks. It discusses the issues around legal, ethical, and practical concerns of using graph-based machine learning and outlines possible harms, business risks, and responsible approaches to applying the AI. Summary of the Problem Identified Payton (2025) analyzed the growing scrutiny of multinational brands to learn where human-rights violations are hidden in expansive global supply chains and demonstrates how AI-specific platforms are rising to meet that demand. The technology in the spotlight is the use of graph-based machine learning, a branch of more advanced data science, which is fed with customs filings, purchase orders, and shipping manifests to create updated knowledge graphs of the global supplier networks. The New-York startup Altana trains representation-learning models to uncover opaque tier-two and tier-three links and statistically mark entities that are associated with forced or child labour. Peer firm FRDM.ai uses natural-language processing and optimisation algorithms to drown out irrelevant noise, match intermediate goods with buyers of their finished products, and estimate the commercial cost of switching to ethical suppliers. The short-term stakeholders are brand-name manufacturers under the looming mandatory due-diligence rules, including the proposed law in South Korea to require disclosure of suppliers with 500 workers or more, investors keeping track of environmental, social, and governance risk, and most exposed workers whose plight is most challenging to quantify. Network Institute for Human Rights and Business and consultancy Anthesis have warned that AI output only has value as the underlying field-based data. Legal and Ethical Issue Analysis A fundamental ethical-legal concern of using graph-based machine-learning systems such as Altana and FRDM.ai to conduct human-rights due diligence is data dependence in the face of opacity. Hot spots of modern-slavery, such as artisan cobalt mines in the DRC or the casual textile shops in Xinjiang, seldom issue purchase orders, bills of lading, or lawsuits. The lack of such datapoints results in a low-risk score, which reports a false negative and can give a false sense of compliance by brands with their statutory obligations under a regime like the EU Corporate Sustainability authority or the 500-employee rule in South Korea. Holzinger, Zatloukal, and Müller, H. (2025) demonstrated that efficacious monitoring practiced using AI requires inclusive datasets, and such monitoring necessitates human oversight as sparsity and reporting bias in impervious places can broadcast the abuses companies see as an advantage of their monitoring methods. The second one is explainability and bias in algorithms. The knowledge-graph embeddings used by Altana and the cost-switch optimisation models used by FRDM.ai are commercially black boxes; however, both the EU AI Act and the accompanying responsible AI frameworks demand providing plausible explanations of decisions that impact fundamental rights. John, Adejumo, and Larbi (2025) illustrated that the outcomes of such supply chains can only be improved when companies put in place safeguards against the risks of adopting AI, such as accountability, fairness, and transparency. These results recommend that no one can delegate legal liability or ethical responsibility to an unreachable algorithm. The strong form of governance should combine AI screening with verifiable field audits, stakeholder consultations, and documented justification that is subject to scrutiny by regulators, investors, and communities on the ground. Descriptive Examples of the Issue Existing examples of graph-based AI due-diligence tools, such as Altana and FRDM.ai, highlight two well-known traps that emerge when algorithms substitute desk-based audits. First, sparseness in opaque markets might result in false negatives. When Altana digests customs document...