Executives have to decide on AI risk controls long before the consequences of AI become visible. Managing AI raises questions of autonomy, learning, and inscrutability, and the risks change as fast as the technology itself: security vulnerabilities, fraud and impersonation, regulatory exposure, vendor dependence, and shifts in market demand. Firms already report these risks. Almost all large US-listed companies now name AI risks in their SEC filings, while only some of them disclose that they deploy AI. Such disclosures show which risks firms see, but they say little about how these risks may develop over the next years or how they interact. Scenario planning helps organizations prepare for uncertain developments by confronting decision makers with several plausible futures at once. Generative AI makes scenarios cheap to produce, but AI-generated scenarios tend to reproduce statistically likely expectations and barely differentiate between time horizons. Time series foundation models can extrapolate risk-related company data separately for each scenario and horizon, which opens a way to ground risk scenarios in data. We still lack knowledge of how such scenarios can inform AI risk governance and how decision makers use them to anticipate and prioritize risks.

This thesis investigates how AI-supported scenario planning can help organizations anticipate AI risks and derive governance responses. Depending on the student's interest, the thesis can focus on one of the following directions:

• how AI risks develop in corporate disclosures, e.g., risk types, sector differences, and the gap between reported risks and reported AI use;
• the design of AI risk scenarios that combine disclosure data, forecasts, and generated narratives;
• how executives judge AI risk scenarios and translate them into governance measures, risk controls, or strategic priorities.

Possible methods include a structured literature review on AI risk and AI governance, the analysis of AI risk disclosures in annual reports (e.g., along established AI risk taxonomies), Design Science Research, expert interviews or Delphi studies, and qualitative content or thematic analysis.

• Close supervision within an ongoing research project at the Institute of Information Systems and Digital Business (IWI-HSG)
• Access to a working futuring prototype and a classified dataset of AI disclosures from SEC filings, including AI risk types and governance responses
• Contact with practitioners and executives from risk management, foresight, and corporate strategy
• Work at the intersection of AI risk, AI governance, strategic foresight, and Information Systems
• The opportunity to publish the results together with the research team

Send me a mail that just briefly describes what you would like to contribute.