Module 2: AI Impacts & Responsible AI · BoK II.A
Organisational harms
Five harm types every organisation deploying AI must price in: reputational, cultural, economic, Acceleration risk and legal. Acceleration risk is the odd one out and the most testable - surprise harm appearing at deployment speed.
Five harm types every organisation deploying AI must price in. This table is lifted almost directly from the course.
| Harm type | Description | Potential impact |
|---|---|---|
| ⭐ Reputational | Loss of trust and credibility after AI failures | Customer loss, share price drop, negative brand perception |
| 🏛️ Cultural | Bias and societal assumptions embedded in AI systems | Reinforced stereotypes, reduced organisational inclusivity |
| 💸 Economic | Costs of litigation and remediation | Significant financial losses, resource allocation strain |
| 🚀 Acceleration | Unanticipated risks from rapid AI processing and complexity | Wider, more severe impacts than traditional technologies |
| ⚖️ Legal | Noncompliance with regulations and laws | Fines, sanctions, operational disruption |
Exam flash: acceleration risk
Acceleration risk is the odd one out and the most testable → it captures harm that emerges because AI moves too fast and is too complex to anticipate. If a scenario describes a surprise harm appearing at deployment speed, that is acceleration.
Key terms - quick answers
What is “Acceleration risk”?
Unanticipated organisational harm from rapid AI processing and complexity, with wider/more severe impacts.