AI as Military Advisor: MIT Tech Review Compiles Six Stories on Defense AI

MIT Technology Review on June 16, 2026 released a subscriber-only eBook compiling six stories on how military organizations are using AI models to inform and make decisions. Written and assembled by journalist James O'Donnell, the collection draws from reporting originally published between April 11, 2025, and April 21, 2026. The stories span AI chatbots advising on targeting, generative AI being used for intelligence gathering, and the Pentagon's plans to let AI companies train models on classified data.

What happened

On June 16, 2026, MIT Technology Review published a subscriber-only eBook titled “How AI is Becoming the Next Military Advisor.” James O’Donnell compiled the package, which bundles six previously published stories updated to reflect developments since their original publication dates. Readers can download it in PDF or ePub format.

The six stories cover a range of military AI topics. According to MIT Technology Review, they include:

• A look at the “new war room” and what matters most in AI right now
• How generative AI is being used to conduct espionage for the US military
• A report on what the publication calls “phase two” of military AI
• A defense official explaining how AI chatbots could be used to help make targeting decisions
• Pentagon plans for AI companies to train models on classified data, as described by a defense official
• How AI is shaping the presentation and perception of the Iran conflict

Why it matters

This is not a think-tank report or a government white paper. It is a working journalist’s compiled account of where military AI actually stands, drawn from over a year of on-the-record reporting. That makes the sourcing more concrete than most coverage of this topic.

The targeting chatbot story and the classified training data story are the most consequential items in the list. Both describe AI being embedded into decisions with direct life-or-death consequences, and both involve named defense officials speaking on record, according to the publication.

For anyone building AI tools or advising clients who work in regulated or sensitive industries, this trajectory is worth tracking. The standards, legal frameworks, and public tolerance for AI-assisted decisions are being set right now, and military use cases tend to set precedents that spread into other institutional contexts over time.

The Iran conflict framing story also deserves attention. Using AI to shape the narrative around an active conflict is a different category of risk than using it to process logistics data. It points to AI being used as a communications and influence tool, not just an operational one.

Our take

The eBook format is a packaging decision, not a research milestone. What is actually interesting here is the reporting underneath it. A defense official going on record about AI chatbots and targeting decisions is a significant data point. That kind of institutional acknowledgment usually comes after the practice is already well established internally.

The classified training data story is the one we would read first. If the Pentagon moves forward with letting AI vendors train on classified material, it creates a procurement and security precedent that will ripple into how government contractors, defense-adjacent tech firms, and eventually commercial cloud providers handle sensitive data and model training agreements.

For most of our clients, none of this is immediate. But the businesses most likely to be affected early are those already selling software to government agencies, healthcare systems, or critical infrastructure operators. Those sectors tend to adopt frameworks from defense contexts faster than others. Watching what rules get attached to this kind of AI use will tell you a lot about where commercial AI governance is heading in two to three years.

One honest caveat: the eBook is subscriber-only. The underlying stories may be available individually through MIT Technology Review’s archive, but the compiled package requires a subscription. That limits how much independent verification most readers can do without paying.

What to do about it

If you work in or near government contracting, defense tech, or any sector with strict data handling rules, read the individual MIT Technology Review stories on AI targeting decisions and classified training data. They are the clearest public record available of where institutional AI governance is heading, and they will give you useful context before your clients or procurement officers start asking questions.

Source: MIT Technology Review