Meta Platforms is testing a prediction-market app with its own employees, according to The Wall Street Journal, as the company examines whether crowd forecasting can become a consumer product without requiring users to wager real money.
The project, reportedly called Arena, allows participants to make forecasts about future events and uses the collective results to reflect what users expect will happen. The internal trial comes before any potential public release, but Meta has not confirmed a launch date or said that the application will be made available.
A play-money model with AI-generated questions
Arena is described as a virtual-currency platform rather than a real-money prediction market. That approach would distinguish it from services such as Polymarket and Kalshi, where participants use real stakes. The design also places Meta’s project within an increasingly visible category of technology products that turn forecasts into a social and interactive experience.
The reported plan would use Meta’s Llama artificial-intelligence model in two ways. Llama would generate questions for users to forecast and provide personalized recommendations about markets or topics. The reported target audience is people aged 18 to 34.
The company is reportedly anticipating at least 100 million monthly active participants, although that figure describes an expected scale rather than an achieved user base. Employee testing is therefore an early stage of a project whose public availability, final features and commercial model remain unconfirmed.
Meta has tried forecasting before
Arena would not be Meta’s first experiment with crowd forecasting. The company previously launched an app called Forecast around 2020-2021, but the service was discontinued in 2022, reportedly because of high operating costs. The new internal test indicates that Meta is revisiting the concept with a different product design and a stronger role for its AI systems.
The Wall Street Journal also reported that Meta is exploring possible partnerships with Polymarket and Kalshi. Those discussions have not been announced as agreements, and an exploration does not constitute a deal. The distinction is important because prediction-market platforms face questions about how their activities should be classified and overseen when real money is involved.
- Product: Arena, a reported internal forecasting app
- Currency: Virtual currency rather than real-money wagers
- AI role: Llama would generate questions and personalize recommendations
- Availability: Employee testing; no public launch date confirmed
For Meta, the project would extend its efforts to use Llama in a consumer-facing service beyond conventional generative-AI interactions. For users, the proposed play-money structure would remove the direct financial stake found on real-money platforms, while still relying on forecasts, rankings or market signals to drive participation. Whether Arena moves beyond internal testing remains unresolved.