Pentagon allocates $600 million to develop advanced agent-based AI

The Pentagon has announced contracts to create an advanced model of "agent AI" that will be able to autonomously plan and execute tasks. This was reported by Militarnyi, citing the Pentagon's Chief Digital and Artificial Intelligence Directorate (CDAO).

It is about artificial intelligence with algorithms for independent decision-making. The total value of the three contracts is $600 million. They were concluded with the companies Anthropic, Google and xAI.

CDAO plans to go beyond generative AI by creating a system that can independently understand a goal, build a plan for its implementation, and adapt to changes to achieve the goal.

The agency notes that agent-based artificial intelligence technology will make it possible to automate tasks that previously required human participation. Automation may include management, analytical, and even combat functions.

The companies' developments will be integrated into the Enterprise LLM Workspace platform , a military solution that allows you to safely work with multiple commercial AI models in a single environment. In the future, the system will be used in the offices of the US Department of Defense.

According to representatives of the department, the involvement of existing commercial solutions will accelerate the implementation of AI not only in the defense sector, but also in intelligence, business and corporate systems. At the same time, the allocated money is a small part of the total funding of companies: in particular, OpenAI attracted $40 billion in investment in March.

The contracts became possible after Google publicly abandoned restrictions on the use of its own AI technologies for military purposes.