{
  "id": 17822512,
  "title": "42AI comes out of stealth with $62.5M and a world model that can ‘predict human behavior’",
  "url": "https://urgent.news/2026/11/02/42ai-comes-out-of-stealth-with-62-5m-and-a-world-model-that-can",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-11-02T16:30:00.000Z",
  "source": {
    "name": "TechCrunch",
    "slug": "techcrunch",
    "url": "https://techcrunch.com/2026/11/02/42ai-comes-out-of-stealth-with-62-5m-and-a-world-model-that-can-predict-human-behavior/"
  },
  "original_language": "en",
  "account": "Israeli AI startup 42AI has emerged from stealth mode with $62.5 million in seed funding. The company's groundbreaking technology, a \"behavioral world model,\" is capable of predicting human behavior even in new situations. 42AI's CEO and co-founder, Amit Gal-Or, explains that this predictive system is currently being tested by a few financial brokerages, where it can help forecast how individual traders will respond to volatile conditions.\n\nAccording to Gal-Or, behavior is not deterministic, but rather constantly probabilistic. People have underlying psychological tendencies that influence their decisions, even when circumstances change. 42AI's goal is to measure and convert these tendencies into data that the model can understand. The startup argues that while a person might change over time, their underlying tendencies will still manifest in certain areas with a certain probability or distribution.\n\nTo achieve this, 42AI examines how individuals react to seemingly unrelated events. For example, when an investor or trader interacts with support or sales, the company looks at sentiment, voice, interaction frequency, and call volume. These data points are then combined with trading activities, market conditions, and other interactions to create a mathematical representation of the individual's traits. This information is then fed into a transformer-based world model that outputs probabilities for possible decisions under different circumstances. The company uses a separate language model to generate explanations for its predictions.\n\n42AI's approach is based on a combination of psychological research, neuroscience, genetics, and insights from experts in psychology and PTSD. The team spent three years developing this technology, studying thousands of individuals through psych assessments and behavioral scenarios, and incorporating government and biobank datasets covering over two million people. The customer datasets have since expanded to cover nearly 30 million people.\n\nIn live deployments, the model analyzes customers' existing records, including transactions and support calls, without requiring personally identifiable information. While the company creates individual behavioral profiles from linked customer activity, the model itself does not know who the person is.\n\n42AI claims its model achieves 98% accuracy in predicting market trades, with an F1 score of 0.8 to 0.85 for predicting trading direction and volume. However, the startup acknowledges that there are natural limitations to the model's accuracy, as it cannot access everything that a person experiences, and its accuracy suffers when important situational context is missing.\n\nThe company is currently working with 10 large brokerages, which are using the technology to manage liquidity, margin requirements, and hedging exposure. 42AI has generated more than \"seven figures\" in revenue, though the exact amount was not disclosed. Customers pay either a flat fee or a percentage of the value gained through the platform. One brokerage reportedly achieved approximately $30 million in additional value through dynamic hedging.\n\n42AI plans to raise additional funds and eventually intends to apply its model to gaming, fraud prevention, and cybersecurity, and eventually help robots and AI agents understand human behavior.",
  "summary": "Here's a question straight out of sci-fi: What if AI could predict what you were going to do before you decided to, even in a completely new situation?",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}