1. Data bias

2. Model bias (gender, racial, cultural)

3. Fairness metrics

4. Debiasing techniques

5. Privacy (differential privacy, federated learning)

6. Environmental cost of large models

7. Responsible AI & NLP