Artificial intelligence is increasingly embedd- -ed in workplace learning platforms through adaptive training systems, skills intelligence tools,learning analytics dashboards, automated assessment,intelligent tutoring,recommendati- -on engines, and predictive workforce develo- -ment systems. These technologies promise personalised learning, improved capability development, efficient compliance training, and evidence-based human resource planning. However, they also intensify the collection, processing, profiling, and secondary use of employee data. In AI-enabled learning environments, the employee is no longer merely a learner but a continuously observed and datafied subject whose behaviours, preferences,performancepatterns,competencies errors, interactions, and development trajectories may be converted into managerial intelligence. This paper examines the ethical boundaries of AI data collection in workplace learning platforms, focusing on consent, privacy,power imbalance, employee autonomy, transparency,purposelimitation,dataminimisati on, algorithmic bias, and organisational accountability. The paper adopts a conceptual and normative research design, synthesising literature from AI ethics, workplace learning, employment relations, data protection, surveillance studies, and algorithmic manage- -ment. It argues that conventional consent models are ethically insufficient in workplace learning because employment relations are characterised by dependency, hierarchy, and unequal bargaining power. Employees may formally consent to data collection while lacking genuine freedom to refuse, withdraw, or contest AI-enabled monitoring. The paper proposes an Ethical Boundary Framework for AI-Enabled Workplace Learning Platforms based on six principles: contextual consent, privacy by design, proportionality, purpose limitation, employee agency, and accountable governance. The framework distinguishes between learning-supportive analytics and surveillanceoriented analytics and recommends clear organisational separation between developme- -ntal learning data and punitive employment decision-making. The paper contributes to the literature by reframing workplace learning platforms as socio-technical governance systems rather than neutral educational tools. It concludes that ethical AI adoption in workplace learning requires a shift from data extraction to data stewardship, where employee dignity, trust, autonomy, and fairness are treated as core conditions of sustainable organisational learning.