China’s military is openly exploring artificial intelligence tools to track undersea targets, analyze satellite trajectories, generate deepfakes, and rapidly fuse battlefield data, according to a comprehensive new study of more than 9,000 procurement notices.
The report “China’s Military AI Wish List,” released in February 2026 by the Center for Security and Emerging Technologies (CSET), analyzed public requests for proposals issued by the People’s Liberation Army (PLA) from 2023 to 2024. What has emerged is not a small number of pilot projects, but a vast catalog of AI ambitions, stretching from the ocean floor to low-Earth orbit.
Visible space and counter space
Among the most impressive findings are explicit demands related to space operations. PLA sought algorithms to detect and classify space objects, determine satellite orbits, and identify anomalies in orbital behavior. Some notices mention systems that can support the detection of space targets and operations under various space weather conditions.
Such language is important. Orbit determination and anomaly detection are widely recognized as components of space domain awareness. In military hands, it could support counterspace capabilities, including tracking satellites and potential targets. The report notes that some space-related requests are unusually brief and specific, suggesting intensive experimentation.
Undersea surveillance challenges Western navies
If orbit is one theater, the depths of the ocean are another. A persistent theme across the dataset is ocean domain awareness and seafloor sensing. Requests included AI-powered acoustic target recognition, sonar data enhancement, and a “global underwater marine environment dynamic analysis system.”
Establishing an environmental baseline—mapping temperature strata, salinity, and background noise—is no ordinary science. Such a baseline is central to detecting subsurface anomalies. By applying machine learning to maritime data, PLA aims to make enemy vessels more visible against acoustic clutter.
Large-scale decision support and data fusion
Beyond sensors, the PLA is investing in AI-enabled decision support systems designed to incorporate open source news, social media, and geospatial data to predict events and assist commanders.
Some requests described systems that can categorize event data by time and location and predict future trajectories.
The short acquisition cycle is outstanding. Many RFPs specify a 3-6 month timeline. This suggests rapid prototyping rather than busy defense bureaucracy.
Deepfakes and cognitive manipulation
The study also identified a desire for “intelligent deepfake systems” that can build multilingual synthetic media libraries and generate manipulated video and audio. Another notice calls for deepfake detection tools, suggesting a dual-track approach: creating and countering.
