Our Methodology
How we generate unbiased AI trust scores
Multi-Model Consensus
We use 4 leading AI models: GPT-4 (OpenAI), Claude (Anthropic), Gemini (Google), and Grok (xAI). Each model independently evaluates companies based on their vast training data, which includes news articles, regulatory databases, financial reports, and public sentiment.
Weighted Scoring
Each AI model provides scores across multiple criteria: regulatory compliance, customer satisfaction, financial stability, transparency, and security practices. We aggregate these into a single 1-5 score using equal weighting across all models.
Bias Prevention
Unlike traditional review platforms, AI models cannot be bribed, incentivized, or manipulated to provide favorable scores. They evaluate based on factual data in their training, providing truly independent assessments.
Regular Updates
We refresh evaluations monthly as AI models are updated with new information. This ensures scores reflect the most current state of each company, including recent news, regulatory changes, and market developments.
Full Transparency
For each company, we show individual scores from each AI model, allowing users to see where opinions converge or differ. We also display the specific questions asked and summarized reasoning from each model.