Pentagon: Slow Adoption of AI Poses Major Risk for the Navy

Pentagon: Slow Adoption of AI, a Major Risk for the Navy

The Pentagon's new AI manual considers slow adoption a greater risk than "imperfect alignment."

The Department of the Navy aims to build an "AI-oriented" fleet. A new strategy outlines how to transform data and AI into battlefield advantages more quickly.

The Department of the Navy, which oversees both the Navy and the Marine Corps, has formally approved a new "Strategy for Arming Data and Artificial Intelligence." Acting Secretary of the Navy, Hung Cao, signed the document, making it immediately effective. The strategy took over a year to develop under the guidance of the department's data and artificial intelligence chief, in collaboration with AI experts across the Navy and Marine Corps.

Cao stated that the strategy would enable the Department of the Navy to "outpace and outfight any adversary" through rapid deployment of data and AI. He described it as a roadmap for building an "AI-oriented" fleet that transforms information into military advantage and allows for faster, more effective decision-making.

The Fastest Learning Force Wins

At the heart of the strategy is the "Bits2Effects Cycle," a five-step framework for digital adaptation. It traces the path from automated collection of military data through transmission, classification, and analysis to its use in actual military decisions and actions. Lessons learned are reintegrated into the cycle, allowing for continuous updates of systems, tactics, and training.

The key metric is the "Mean Time to Effect," or MTTE. It measures the time from when new data is captured to when it produces a concrete military response or adaptation. The shorter this window, the more quickly a force can react and adjust. In a prolonged conflict with multiple learning cycles, the force that learns and adapts the fastest will dominate, according to the strategic document.

The announcement establishes six objectives:

• Accelerate operational deployment of AI
• Improve data availability and usability
• Expand technical infrastructure
• Streamline approval processes
• Strengthen the culture of data and AI among personnel
• Deepen collaboration with industry, academia, government agencies, and allies

Many of these measures are expected to be implemented by the first quarter of fiscal year 2027, which ends in December 2026. By the end of fiscal year 2029, the number of qualified data engineers, data scientists, and AI and machine learning engineers is expected to double.

Moving Too Slowly is Riskier than "Imperfect Alignment"

The strategy calls for running large language models and agentic AI directly on warships and with Marine Corps expeditionary units. These systems must operate even when communications are jammed or cut off. Service members would be able to create their own applications on top of this. An "AI War Council" would prioritize use cases, coordinate resources, and pre-approve changes during wartime regarding data sharing, classification, and deployment rules.

The strategic document adopts a particularly broad compromise compared to the Department of Defense's more general AI strategy: the risks of moving too slowly outweigh the risks of "imperfect alignment" in these systems. This passage is framed within a "Wartime Approach." The department aims to treat risk assessments and organizational hurdles as if the country were already at war, making decisions that favor speed.

AI is Already a Reality on the Battlefield for the U.S. Army

The Navy's strategy is part of a broader transformation of AI across the U.S. armed forces. Business Insider reports that GenAI.mil, the central platform where personnel and employees of the Department of Defense can use generative AI, reached 1.5 million daily users in June 2026, up from 80,000 at its launch in December 2025. Uses range from routine office tasks to military planning and combat operations.

The Army is testing AI in a "Next Generation Command and Control" system to process large volumes of data more quickly and help soldiers build situational awareness and make decisions. A Navy AI program reportedly reduced a submarine planning task from 160 hours to ten minutes.

The reality of these applications became clear during the war against Iran. The U.S. military reportedly used the Claude language model from Anthropic for target analysis and strike planning. This deployment is politically charged.

The Trump administration had excluded Anthropic from government systems after the company insisted on restrictions regarding fully autonomous weapons and domestic mass surveillance. Shortly after, OpenAI struck a deal with the Pentagon to run its models on classified networks. OpenAI cites similar red lines but relies on contractual and technical assurances rather than strict political requirements. The Navy's new strategy is likely to increase military demand for powerful language models and AI agents.

A Global AI Arms Race

The AI arms race is unfolding on a large scale worldwide. China is rapidly pushing for military AI adoption. Researchers from Georgetown University have analyzed thousands of public procurement requests from the People's Liberation Army of China. The documents show that Beijing is testing AI systems for unmanned combat vehicles, cyber defense, tracking ships, target acquisition on land, at sea, and in space, as well as deepfake-driven disinformation.

NATO is also operationally using AI. French Admiral Pierre Vandier, responsible for NATO's digital transformation, stated that alliance members are using AI to track Russia's ghost tanker fleet. Israel has spent years deploying AI to sift through the flow of intercepted intelligence data before its war against Iran.

On the U.S. side, the Pentagon is heavily investing in integrating commercial AI and plans to go further by allowing AI companies to train army-specific model versions on classified data. This would represent a qualitative leap. Sensitive intelligence would be integrated directly into the models.

Cybersecurity, Where the Stakes are Highest

The pace is rapidly accelerating in cybersecurity. The parallels with nuclear escalation are no longer just a metaphor. Zhou Hongyi, founder of the Chinese cybersecurity company Qihoo 360, has explicitly made this comparison. He argued that the ability of AI models like Claude Mythos from Anthropic to autonomously find vulnerabilities and build attack chains equates to "cyber nuclear weapons of the AI era."

The urgency behind this rhetoric is rooted in measurable technical advancements. The UK's AI Security Institute has revised its estimate of how quickly AI's cyber capabilities are doubling, adjusting it upward twice in a few months. The U.S. government now considers these models strategic assets and initially blocked Anthropic from publicly launching its AI model Fable 5.

Zhou described Fable 5 as a "civilian and neutralized version of Mythos," Anthropic's top-performing cybersecurity model, and suggested that the U.S. fears foreign actors circumventing the system to achieve Mythos-level capabilities. "This is what the U.S. government finds most intolerable. It must ensure that it alone possesses this capability, forming an absolute monopoly on this strategic asset," Zhou stated.

The European Union is stuck on the sidelines, dependent on the goodwill of major American tech companies as comparable European products do not exist.