#Bias Detection

Evaluating Feature Steering: Anthropic's Exploration of Mitigating Social Bias in AI

<p>Anthropic's latest study explores the use of feature steering to mitigate social biases in their Claude 3 Sonnet model.</p> <p>Researchers identified a "sweet spot" for steering features to reduce bias without impairing the model's capabilities.</p>

Mona Announces New AI Fairness Feature for Its Monitoring Platform