Two former OpenAI founders have built a tool designed to measure how well artificial intelligence models actually know who you are, according to a report from Startup Fortune.
The premise gets at a question that has quietly become central to the way people use AI. As chatbots and assistants take on more personal roles — drafting emails, answering questions, making recommendations — how much they genuinely "understand" about an individual user shapes how useful, and how trustworthy, their answers turn out to be. Startup Fortune frames the new tool as a way to put a number on that, gauging how accurately a given model represents a specific person.
The involvement of ex-OpenAI founders is notable in itself. OpenAI, the maker of ChatGPT, sits at the center of the current AI boom, and people who helped build it carry both technical credibility and a close-up view of how these systems handle personal information. According to Startup Fortune, this team has now turned that experience toward measurement rather than model-building — creating an instrument to evaluate AI behavior instead of another assistant.
The available source is a single headline-level report, so finer details remain limited: the names of the founders, the tool's exact methodology, who can use it, and what the early results show are not specified in the material at hand. Those gaps matter, because a measurement tool is only as meaningful as the way it defines and tests "knowing" someone.
Still, the direction is clear. Much of the debate around AI has focused on raw capability — how well models write, code, or reason. A tool aimed specifically at how well a model knows an individual points to a different and increasingly important axis: personalization and accuracy at the level of the person.
Why it matters: as AI assistants weave deeper into daily life, having an independent way to measure what they actually know about us could help users and companies judge when to trust the machine — and when not to.