China’s Military Is Now Using AI to Plan Strike Operations

The People’s Liberation Army (PLA) Air Force now uses an AI system that draws up plans for large-scale strike operations. Instead of being released with great fanfare, the system appeared in a CCTV documentary released in early August to mark the PLA’s 99th anniversary. The system is described as prioritizing targets from a pool of hundreds, coordinating dozens of formations, and assigning tasks to more than 100 tactical units. According to the documentary, it has already been used in multiple missions.

The timing lines up neatly with Xi Jinping’s instruction in late July that the PLA must strengthen the military application of unmanned intelligent technologies and gradually build an intelligent military system. As China has been pursuing this very goal for years, Xi’s remarks are not new. The level of emphasis, however, is high: Beijing wants steady progress toward the modernization goals set for 2027, which marks the 100th anniversary of the PLA’s founding.

The AI system is described to handle a volume of decisions that would otherwise overwhelm human planners as, so goes the argument, traditional calculation methods cannot keep up when large numbers of aircraft and targets move at high speed. This indeed is a fair assessment and is far from being unique to China. 

Unlike autonomous weapon systems, the software does not select and engage targets directly. Its significance lies higher in the command chain. It helps plan air campaigns by recommending how aircraft, electronic warfare assets, and supporting capabilities should be combined before commanders authorize execution. In other words the system is described as a support tool that helps commanders and pilots – but humans remain responsible for the final decisions.

As is often the case with great powers and military AI, there is a catch in China’s posture and the potential operation realities. On the surface, Beijing has long maintained that artificial intelligence should support human judgment rather than replace it. China’s statements on lethal autonomous weapon systems at the Convention on Certain Conventional Weapons emphasize that AI should support rather than replace human decision-makers.

On the other side of the AI coin, however, the system’s apparent ability to compress hours of complex campaign planning into minutes risks reducing human authority to barely more than a formal rubber stamp, thus creating a sharp tension between China’s diplomatic posture and its operational realities.

Beijing’s logic is understandable. Decision superiority has become an increasingly important determinant of military effectiveness alongside the quality and quantity of platforms and weapons. Compressing the planning cycle allows commanders to adapt faster to changing battlefield conditions and exploit fleeting opportunities. The problem is that even if humans remain formally in control, AI can alter the pace of conflict enough to reshape how that control is exercised.

When planning happens in minutes, the room for independent human assessment shrinks together with the time frame. Even for pre-validated targets the system works with, judging whether it is military or civilian, whether the expected civilian harm is proportionate, and whether the attack is necessary all require time. 

Under high tempo operations, like executing a coordinated air offensive across the Taiwan Strait, the human role can become little more than rapid approval of options the AI has already ranked. Indeed, this may already be happening in Israel’s strikes on Gaza. A source told reporters, “We prepare the targets automatically … We work quickly and there is no time to delve deep into the target.”

Under such circumstances, the risk of automation bias rises. Accountability grows harder to pin down. If a strike based on an AI-generated plan produces unlawful effects, whom does responsibility sit with exactly? With the commander who accepted the plan? Or with those who designed the system that produced it? Ultimately, keeping a human in the loop satisfies diplomatic requirements, but in practice risks reducing commanders to a convenient legal scapegoat for algorithmic failures

These issues carry weight across the Indo-Pacific and beyond. A system built for fast, coordinated campaigns raises the speed at which China could generate and execute complex strike packages in a South China Sea scenario, for example. Crisis management tools like direct military hotlines or diplomatic signaling that depend on slower human judgment and dialogue may well end up lagging dangerously behind. Consequently, countries in AUKUS and the Quad can no longer rely on traditional de-escalation but must build defense strategies capable of absorbing and countering potential algorithmic shocks.

These strategic implications are far reaching and the system offers a revealing glimpse into how AI can reshape the conditions under which humans may well be making decisions about the use of force. Meaningful human control requires far more than a formal rule that a person stays in the loop, and all is in the green. It needs interfaces, review steps, and training that still allow real judgment when time is short. Compliance with the law of armed conflict works better when those requirements are built into the system from the beginning rather than added later as a quasi-afterthought.

It is important to highlight that the planning system is far from being an isolated project; it fits a broad Chinese pattern. Recent disclosures span AI integration across logistics, intelligence analysis, and battlefield management, pointing to Beijing’s vision for an end-to-end algorithmic command ecosystem. Viewed together, this strategy maintains human oversight in name while embedding machine speed into every stage of military operations. In effect, the service member remains in the loop not to exercise operational control, but to co-sign the machine’s momentum. 

As systems like this one move from demonstration into regular use, Beijing’s statements about human control are bound to face closer examination. Yet without mechanisms to inspect software or verify decision loops, diplomatic pushback will remain stuck somewhere between meager and nonexistent. Meanwhile the actual, for now hard to assess, quality of oversight that PLA branches install in their planning tools will shape both the character of future operations and the chances for workable rules on military AI globally. 

That is to say, the AI arms race just got an algorithmic nitro boost. As systems like this will inevitably proliferate, the fear of falling behind will drive other nations to field their own AI planners, ensuring that machine-speed warfare becomes the global standard. In this brand new, AI driven era, the ever increasing risk is not just that algorithms might fail, but that humans will no longer have the time or agency to stop them.