How bias creeps into the AI designed to detect toxicity

Bias can creep into the AI models used to determine whether comments are 'toxic' versus 'non-toxic,' leading to harmful mistakes.