Responsible AI Is Shifting from Legal Defence to a Condition for Growth and Trust

Harvard Business Review published an article on July 21 by Michael Wade of IMD and Jochen Wirtz of the National University of Singapore. They argue that responsibility for AI is moving from a defensive legal agenda into business strategy. “Trust is becoming a decisive competitive advantage,” they write. Customers, employees, and business partners judge not only system performance, but also whether a company handles data and decisions predictably. The authors describe a basic tension. The same capabilities that personalise a service, automate support, and generate operational insight can also be used for surveillance, manipulation, or discrimination. Technological benefit cannot therefore be separated from the way a system is used. A company that measures only speed and cost may overlook harm that later appears as a fine, lawsuit, customer loss, or restricted market access. The article gives three specific examples. Robotic-toy maker Apitor was penalised in 2025 for collecting children’s location data and passing it to a third party in China. Sirius XM faces a lawsuit alleging that a recruitment tool screened out Black applicants through proxy variables, such as postal code and school, before a person reviewed the résumé. Australian authorities also found, according to HBR, that two retail chains owned by Wesfarmers used facial recognition without adequately informing customers. These cases share a governance-design failure: the company did not fully consider what data the system collected, why it used it, or whether the affected person could understand or challenge the process. Formal consent or general terms and conditions may therefore be insufficient. Wade and Wirtz use the term corporate digital responsibility. It should cover the entire product life cycle: data selection, development, model procurement, testing, market launch, incident monitoring, and service termination. In practice, this means lawyers, security experts, and commercial representatives participate in design before the product is finished. Late review often identifies a problem when fixing it is expensive or commercially impossible. The business case for responsibility arises in three areas. A company with a trustworthy product can more easily obtain sensitive data, sign a regulated customer, and preserve its brand during an incident. It must, however, measure the advantage. Alongside revenue and costs, it should track complaints, erroneous interventions, differences between user groups, time to remedy, and the share of suppliers that met control requirements. Responsible AI should not mean eliminating all risk. It means a conscious decision about which risk the company accepts, who owns it, and what threshold triggers intervention. “The same capabilities can personalise and discriminate,” the authors warn. Management must reflect this dual nature in pricing, development, and contracts. Trust becomes an economic asset only when the company can prove that its promises match actual operations. Terms explained Corporate digital responsibility: Principles for the safe, fair, and transparent use of data and digital technologies. Proxy variable: Data that indirectly stands in for a sensitive trait, such as postal code as a signal of background. Facial recognition: Automated comparison of human facial features with stored images or data patterns. System life cycle: The period from design and development through operation to retirement and removal.

Responsible AI Is Shifting from Legal Defence to a Condition for Growth and Trust

The full article is available to users registered at Hard Skills