A new study from AI Forensics reveals concerning biases in Apple Intelligence when processing ambiguous queries. Researchers analyzed over 10,000 notification summaries and found the system defaults to treating White people as the baseline while applying gender stereotypes when no gender is specified.

When ethnicity is mentioned in messages, Apple's AI tends to reflect that information in summaries. However, the system largely ignores racial context for Caucasian individuals, treating them as the default category. Gender biases also emerged, with the AI making assumptions about gender when none was provided in the original message.

These findings raise questions about the training data and algorithms behind Apple's AI features. The biases appear most prominently when queries lack specific details about race or gender, suggesting the system relies on problematic assumptions to fill in gaps. As Apple continues to expand its AI capabilities across devices, addressing these biases will be critical for ensuring fair and accurate results for all users.