The Pentagon opens the door to a greater role for AI in military targeting

By Seeking Alpha (via Bloomberg) · June 25, 2026.

The U.S. Department of Defense has revised its classified doctrine on military targeting to explicitly include a far more active role for artificial intelligence in combat decisions, according to a report by Bloomberg News on Thursday. The Seeking Alpha article —partially behind a paywall— reports that the revised guidance has already been approved, although the full details of the document remain classified.

The doctrinal change marks a significant milestone in the Pentagon-driven process of automating military operations. Until now, official Department of Defense policy held that a human had to maintain meaningful control over the use of lethal force. The update, according to the available information, does not remove that human responsibility —commanders remain legally responsible for AI-enabled use-of-force decisions—, but it does expand the space in which autonomous systems can operate and recommend actions more directly.

From a market standpoint, Seeking Alpha highlights three benchmark ETFs in the aerospace and defense sector that would be directly affected by this doctrinal shift: the Invesco Aerospace & Defense ETF (PPA), the iShares U.S. Aerospace & Defense ETF (ITA) and the SPDR S&P Aerospace & Defense ETF (XAR). The logic is straightforward: greater integration of AI in military operations means higher-value contracts for companies specializing in autonomous systems, decision-support software and platforms with advanced sensors.

**Implications for defense sector demand**

According to the "Quick Insights" published in the article itself, the Pentagon's new doctrine is expected to increase long-term demand for AI-enabled systems in the defense sector. The direct beneficiaries would be three categories of companies: (1) traditional defense contractors accelerating their transformation toward autonomous platforms; (2) developers of software specialized in secure artificial intelligence for military environments; and (3) cloud infrastructure providers with government security certifications, needed to run classified AI models.

In general, as sector context, the global military AI market was already in rapid expansion even before this announcement, with multiple Department of Defense programs —such as the JADC2 (Joint All-Domain Command and Control) program— aimed precisely at connecting sensors, weapons systems and command centers through real-time AI networks.

**Specific opportunities for military AI companies**

The article identifies four technology areas with the greatest growth potential in this new environment: secure and auditable AI models for military use, autonomous platforms (ground, air and naval), advanced sensors for target reconnaissance and designation, and decision-support software that helps commanders act faster with more information.

As sector context, companies such as Palantir Technologies, Anduril Industries (still privately held), Shield AI and L3Harris have positioned a significant part of their business precisely in this type of solution. The major traditional contractors —Lockheed Martin, Raytheon (RTX), Northrop Grumman, General Dynamics— also compete actively for these contracts, although their transformation toward software and AI is more gradual.

**Ethical, legal and regulatory risks**

The article itself acknowledges that adopting AI for combat decisions carries elevated legal and ethical risks. The updated doctrine maintains that commanders are ultimately responsible for AI-enabled use-of-force decisions, which creates a structural tension: if the autonomous system recommends —or executes— an action in milliseconds, human responsibility becomes difficult to exercise in practice.

In the international regulatory arena, the debate over lethal autonomous weapons systems (LAWS) has been active for years at the United Nations without a binding treaty being reached. The Pentagon's new stance could reignite that debate and generate diplomatic pressure on the United States from European allies that have adopted more cautious positions. From the perspective of the European Union's AI Act, although this is a sovereign U.S. regulatory act not subject to European rules, the precedent it sets may influence how the EU regulates exports of dual-use (civilian and military) AI technology to third countries.

**Agentic perspective**

From an agentic AI standpoint, this doctrinal update is especially significant. The most advanced military targeting systems are not simple image classifiers: they are agents that process information from multiple sensors, evaluate options, prioritize targets and recommend —or potentially execute— actions without human intervention at every step. The fact that the Pentagon formalizes a greater role for these agents in its classified doctrine means that the requirements for reliability, explainability and control of agentic AI systems in high-risk contexts will become a contractual priority, not just an academic one.

This also opens the door to new technical requirements: AI systems that can justify their recommendations in an auditable way, robust interruption mechanisms (kill switches), and architectures that ensure the chain of legal responsibility remains intact even when the speed of decision-making exceeds the human capacity for real-time oversight.

**Note on sources**

The original Seeking Alpha article is partially protected by a paywall. The available information comes from the headline, the published opening paragraph and the "Quick Insights" section accessible without a subscription. The primary source of the news is Bloomberg News. The full details of the revised doctrine remain classified.