In the Weights turns model recall into an AI vanity search

TechCrunch reported on In the Weights, a new AI-centric vanity-search tool created by Thomas Dimson and Joey Flynn after they left OpenAI. The site queries multiple models, including GPT, Claude, Gemini, Grok, and Llama variants, to see how well they can recall a person's identity without using web search. It clusters the models' answers, assigns a strength score, and highlights hallucinated or conflicting descriptions. The tool is playful, but the underlying idea is serious: as more people are discovered through AI answers rather than search results, model memory, training-data prominence, and hallucinated identity claims become a new reputation surface.

Key details: Published June 20, 2026 at 19:41 UTC, The site was built by Thomas Dimson and Joey Flynn, formerly of Global Illumination and OpenAI, It queries several AI model families to score how strongly they recall a person, The results expose both model memorization and hallucinated identity descriptions.

Why it matters: The project makes a fuzzy problem visible: people and companies are starting to care not just how they rank in search, but how models remember or misremember them.