Celebrities like Macaulay Culkin and Luciano Pavarotti currently rank highest on a new tool measuring how strongly AI models recall information about a person without external search, revealing a surprising dimension of digital fame. Former OpenAI employees Thomas Dimson and Joey Flynn launched Intheweights, a free website enabling users to examine their digital footprint in AI training data—effectively an AI-centric vanity search engine for 2026.
Yet, the public assumes AI models are comprehensive, accurate knowledge bases. 'In the Weights' exposes a stark reality: AI's internal memory of individuals is often selective, incomplete, and prone to fabrication, according to Zamin Uz and The Tech Buzz. Organizations relying on AI for biographical data are building on quicksand. AI's 'knowledge' is a flawed, hallucination-prone construct, not a reliable source of truth.
Therefore, individuals and organizations must now actively consider their 'AI-native' digital footprint as a critical aspect of their public image, or risk significant misrepresentation by increasingly influential AI systems.
How AI Remembers You
The 'In the Weights' project, from former OpenAI employees Thomas Dimson and Joey Flynn, determines a person's presence within Large Language Models' (LLMs) internal parameters ('weights'), according to Zamin.uz. The tool measures how well AI models recall individuals without external search, assigning a 'strength score' based on stored knowledge, reports Startup Fortune. It queries models like Grok, Gemini, GPT, Claude, and Llama with "Who is this person?" then analyzes responses for that score, per Zamin.uz. This process offers a direct look into AI's pre-trained memory, revealing not what it can find, but what it inherently 'knows' about a person.
The Surprising Flaws and New Metrics of AI Memory
'In the Weights' exposes AI 'hallucinations'—confusing individuals or fabricating facts, according to Zamin.uz, directly contradicting AI's perceived authority; its internal memory is selective and error-prone. The high ranking of figures like Macaulay Culkin and Luciano Pavarotti, per Zamin.uz, highlights a critical flaw: AI often prioritizes historical data over current relevance, meaning AI-shaped public perception risks becoming a distorted echo of the past, not a reflection of the present. The tool not only reveals AI's internal memory unreliability but also establishes a novel metric for digital prominence.
