PlumeSense AI enables real-time monitoring of underground CO₂ storage by replacing traditional geomechanical simulations with a fast, AI-powered surrogate model. Trained on high-resolution data, it predicts plume dynamics in milliseconds with under 10% error and is around 10,000 times faster than conventional methods. This allows minute-scale leak detection, reduces the need for dense sensor networks, and lowers operational costs. Validated on real-world data, the system offers a safer, scalable approach to carbon capture and storage, supporting public trust and regulatory oversight