A recent study by Salesforce AI Research’s Pranav Narayanan Venkit reveals that nearly one-third of generative AI tools display bias and lack reliable sourcing. For OpenAI's GPT-4.5, 47% of claims lacked support, with 97.5% of those from Perplexity’s deep research agent also unverified. Evaluated through the DeepTrace metrics, the findings highlight significant issues in AI response accuracy and bias, stressing the need for improvements in how these systems process information.