Abstract
The growing adoption of Artificial Intelligence (AI) has heightened concerns about the need to raise awareness of AI’s potential risks. Although several studies have explored AI-related risks, a lack of practical tools to support comprehensive and accessible risk assessment remains. To address this gap, we present an approach that assists in developing a practical solution. The proposed method can be employed to build probabilistic models derived from news reports on incidents involving AI technologies. It aligns with key requirements identified in the literature on AI risk assessment and enables efficient data retrieval and analysis. These capabilities can then be used to support quantitative risk assessment. The feasibility and effectiveness of the approach are validated through a proof-of-concept implementation.