Carolyn Geason-Beissel from MIT SMR highlights a critical, often overlooked issue in AI governance within Fortune 500 companies. While every company claims to govern AI, few can identify who has the authority to shut down an AI model causing harm. This gap reveals a structural problem: governance teams can monitor and flag issues but lack the power to act decisively. Most decision authority often resides with product teams prioritizing revenue, not accountability. Adobe’s solution was a federated governance model, with AI system owners and a steering committee reporting independently from product teams, ensuring clear accountability. With regulations like the EU AI Act demanding documented decision-making and accountability, companies must build governance structures that empower individuals with real authority to intervene, not just oversee. The future of AI governance requires more than tools and policies—it demands human accountability backed by organizational power.
Back
