A research team from Peking University, led by Wang Heyuan and Wang Kai at the Institute for Carbon Neutrality (ICN), has used advanced AI models to reveal that the global land carbon sink has been halved in 2024. The study, titled "AI-tracked halving of global land carbon sink in 2024," was published in Science Bulletin and highlights a dramatic decline in the Earth’s ability to absorb carbon dioxide from the atmosphere due to an abrupt and extreme rise in global temperatures. The researchers employed machine learning algorithms trained on decades of satellite data, climate observations, and ecosystem measurements to track changes in carbon uptake across terrestrial ecosystems. Their findings indicate that the land carbon sink—once responsible for absorbing nearly a third of human-caused CO₂ emissions—now sequesters only about half as much carbon as it did in previous years. The sharp reduction is attributed to extreme heatwaves, prolonged droughts, and widespread wildfires, all intensified by climate change. These conditions have stressed forests, grasslands, and other land-based carbon reservoirs, turning some regions from carbon sinks into carbon sources. For example, large parts of the Amazon, boreal forests in Siberia, and savannas in Africa have shown signs of reduced photosynthetic activity and increased respiration, leading to net carbon release. The AI model’s ability to detect and quantify these shifts in near real time marks a significant advancement in climate monitoring. Unlike traditional methods that rely on sparse ground measurements and delayed data processing, the AI system integrates multiple data streams to provide a more accurate, dynamic, and timely picture of the global carbon cycle. The study underscores the urgency of climate action, as the weakening of the land carbon sink could accelerate global warming in a feedback loop. The researchers warn that without rapid emissions reductions and stronger ecosystem protection, the planet’s natural capacity to buffer climate change may continue to erode. The findings are expected to inform future climate policy, carbon accounting, and international climate agreements, emphasizing the need to integrate AI-driven insights into global environmental monitoring systems.