Walking the Walk of AI Ethics: Organizational Challenges and the Individualization of Risk among Ethics Entrepreneurs

Ethics implementation lags behind ethics statements: The research found a significant disconnect between what technology companies say about AI ethics and what actually happens in practice, revealing that corporate ethics policies often exist on paper without translating into real changes in product development.

Organizational structures actively undermine ethics work: AI ethics workers face three specific institutional barriers—pressure to prioritize product launches over ethics considerations, difficulty measuring ethics outcomes in metrics-driven environments, and frequent team reorganizations that disrupt continuity—making it nearly impossible to institutionalize ethical practices.

Individual workers bear disproportionate personal risk: The study shows that employees who raise ethics concerns, particularly those from marginalized backgrounds, take on significant personal risk to do so, suggesting that companies have not created safe structures for ethics accountability and instead rely on individual moral courage rather than systemic safeguards.

"Ethics entrepreneurs" are working against the system, not with it: Rather than ethics being integrated into normal business operations, individual workers must act as entrepreneurs to push ethics initiatives forward, indicating that responsible AI requires fundamental organizational change—not just hiring ethics staff.

Current corporate structures may be incompatible with genuine AI ethics: The research suggests that technology companies' existing incentive systems, metrics-based goal-setting, and product-launch timelines are structurally misaligned with doing ethics work, raising questions about whether meaningful AI ethics is possible without deeper organizational transformation.