Question

A financial services company wants to fine-tune a foundation model using its internal dataset, which includes customer account numbers and transaction histories. The security team is concerned that the model could later reproduce sensitive details in its responses to other users. Which responsible AI concern does this situation most directly raise?

Accepted Answer

Data privacy, because sensitive personal information used in training could be memorized and leaked in generated outputs

Answer

Bias, because financial data may over-represent certain customer demographics

Answer

Transparency, because users cannot see which model version answered their question

Answer

Hallucination, because the model may invent account numbers that do not exist