Pollution does not stop at the boundary between ports and cities, yet calculation-based methods cannot determine whether it originates from vessels, traffic, factories, or background sources. This attribution gap blurs responsibility and delays action to protect port workers and residents. Emission Guardian AI is a shore-based platform that measures, verifies, and predicts emissions in real time. Vision AI, thermal imaging, sensors, AIS, weather, and wind-backtracking link smoke and plume movement to vessels. Models trained on over 80,000 carbon and emission measurement records combine them with shore measurements, separating vessel-attributable pollution from urban sources and improving reliability by over 30% versus sensor-only analysis. Before vessels arrive, the platform identifies high-risk ships and hotspots, sharing alerts, evidence, recommendations, and reports with port authorities, local governments, environmental agencies, and health officials. It turns the port-city attribution gap into a shared, evidence-based response system for earlier intervention, cleaner air, and safer communities.