ML model risks · Question 26 of 60
Against which risks does red teaming help?
3 answers are correct. Medium
Show answer
Answer
- A Harmful or undesirable outputs such as racist views or dangerous guidance
- C Security and safety vulnerabilities
- D Privacy violations
Explanation
Red teaming is the mitigation for security and safety vulnerabilities, privacy violations and harmful or undesirable outputs. Overfitting is addressed by testing for overfitting, and performance efficiency by performance testing.