Predicting Human Behavior With a Behavioral World Model

Israeli startup 42AI has emerged from stealth with $62.5 million in funding, introducing a system that claims to predict what individuals will do before they make a decision. Rather than extrapolating past actions, the company built a behavioral world model designed to measure internal psychological tendencies and convert them into probabilistic data.

The startup spent three years developing the approach by combining psychological research, neuroscience, and genetics, initially studying thousands of people before expanding its training sets to millions of records. In live deployments, the model analyzes customer records, including support calls, onboarding interactions, and transactions, without using personally identifiable information.

Financial brokerages are currently using the system to anticipate how traders will respond to volatile market conditions, helping firms manage liquidity and hedging exposure. While founders Amit Gal-Or and Ehud Ezra argue that behavior is probabilistic rather than deterministic, they acknowledge the system faces a natural ceiling when missing critical situational context like advice from a colleague.

Our read is that transforming subjective human motivations into mathematical traits turns unpredictable customer psychology into a quantifiable asset for financial firms.

That capability creates an acute tension regarding how predictive profiles might be leveraged against the individuals being analyzed.

When asked what prevents brokerages from using this foresight to gain an unfair advantage, the founders pointed to regulations, monitoring, and an internal ethics committee while maintaining that the technology itself forms part of the solution.

As 42AI looks beyond finance toward gaming, cybersecurity, and eventually teaching AI agents how humans think, the core question remains whether algorithmic anticipation of human choice can ever truly be governed.