Pentagon Deploys Agentic-AI Targeting Tool Backed by Palantir as Governance Concerns Mount

The Pentagon has announced a new agentic-AI system called Agent Network that will continuously scan intelligence feeds and operational data to provide U.S. military commanders with targeting options "within seconds," raising both enthusiasm and concern inside and outside the Defense Department.

The tool employs AI "agents"—software entities that perform tasks autonomously on behalf of a user, such as running scheduled searches or processing data streams—to "continuously scan defense intelligence and operational systems, translating findings into clearly presented options," according to a Pentagon press release.

Officials were explicit that the system does not make decisions on its own. "Agent Network does not autonomously select or strike targets; it ensures commanders remain in charge of every decision," the release stated.

Agent Network is one of seven "pace-setting" projects first unveiled in January alongside a new Pentagon AI strategy. Key contractors on the effort include Lumbra and Palantir, which already handles significant targeting analysis through its Maven Smart Systems contract.

Despite the official optimism, experts are divided on how much the technology can currently deliver.

Vishal Sikka, a former CEO of SAP, argued last July that "tasks that AI agents are instructed to perform can clearly have computational complexity beyond" what current large language model architectures can handle. Citing the Time-Hierarchy Theorem, Sikka noted that transformer models approach difficult and simple tasks using the same mechanical formula, processing only a limited number of operations per token. "Despite their obvious power and applicability in various domains, extreme care must be used before applying LLMs to problems or use cases that require accuracy, or solving problems of non-trivial complexity," he wrote.

Others push back on that skepticism. Illia Pashkov, founder of SINT Labs and editor of The Agent Times, said the technology has moved well past the prototype stage. "Agentic AI quietly stopped being a demo this year," Pashkov said. "It's drafting code, clearing support queues, grinding through back-office work in finance and healthcare, and now it's reading intelligence. The speed is not hype. I've watched these systems compress weeks of analyst work into an afternoon."

Pashkov nonetheless warned that speed and confidence can be a dangerous combination. He cited a private-sector case in which a company's AI agent deleted a live production database. "The danger was never a dumb agent; it's a confident one running without a leash, a logbook, or a human who owns the call," he said.

Inside the Defense Department, that governance challenge is already drawing attention. A DOD intelligence security official not directly affiliated with the Agent Network program described an atmosphere of broad enthusiasm for agent systems across multiple Pentagon offices. "There are so many opportunities to leverage the DOD Enterprise capabilities and allow people to build their own agents," the official said. But the same official acknowledged that tracking how every agent is performing is a serious challenge, adding that governing all of them "will be nearly impossible."

The announcement comes as agentic-AI deployment accelerates across both public and private sectors, with organizations betting that autonomous software can process information at a scale and speed no human analyst could match. For the military, the stakes of that bet are considerably higher than in a corporate back office.

What happens next will likely depend on how rigorously the Pentagon enforces the human-in-the-loop constraints it has publicly committed to—and whether the governance frameworks can keep pace with the rate at which new agent systems are being stood up across the department.