Question

A retail company wants to fine-tune a generative AI model using internal documents that contain employee names, salaries, and home addresses. The legal team raises concerns before the project proceeds. Which responsible AI concern is MOST directly at issue, and what is the best first step?

Accepted Answer

Privacy — remove or anonymize the personally identifiable information from the training data before fine-tuning

Answer

Transparency — publish a model card explaining how the model was built before doing anything else

Answer

Fairness — rebalance the dataset so all employee groups are equally represented

Answer

Security — encrypt the network connection used to upload the documents to the training environment