Chile's Army adds AI ethics to its officer training, before having any doctrine on the subject

By Zendoric · September 10, 2026.

The Chilean Army's Military Polytechnic Academy (ACAPOMIL) organized, as part of its centenary programme, a talk by Mois Navon titled "Artificial Intelligence: an introduction for non-specialists". Navon is an engineer and philosopher and was chief engineer at Mobileye, the Israeli computer vision company for autonomous vehicles acquired by Intel; he holds six patents in image processing and has worked on defence, aeronautics and medical technology projects, according to the Chilean Army itself. The session was attended by students from the military engineering and General Staff courses, as well as members of military units and of the academic world.

According to the institutional account, the content covered the evolution of AI, autonomous systems and their ethical implications, with a focus on the relationship between digital transformation and human dignity, and applying those dilemmas to concrete military scenarios. Colonel Alex Hellman, director of ACAPOMIL, summed it up as an invitation to "think of technology not as an end, but as a means" to strengthen institutional capabilities and to train professionals with "judgment, prudence and strategic vision". Navon himself noted that it was the first time he had addressed a military audience, having previously done so at universities and civilian institutions.

In itself, it is a minor event: an outreach talk, with no product announcements, no new doctrine, no deployment figures. But it merits attention because it fits a broader trend we have already been tracking: the ethics and governance of military AI is ceasing to be the exclusive terrain of Washington, Brussels or Geneva —where the United Nations has spent years discussing, without any binding agreement, how to regulate lethal autonomous weapons systems— and is beginning to filter into officer training in countries that are not on the front line of that technological race. Chile is not developing autonomous weapons, nor does the article suggest it; what it is doing is the cheapest and perhaps most useful thing a mid-sized military institution can do: teach people to think about these systems before having to make decisions about them.

That sequence —forming judgment before drafting doctrine— is, in general, what has been missing from much of the civilian adoption of AI: organizations buy the tool and then improvise the rules for using it. That a military academy should reverse the order, inviting an engineer with a track record in applied ethics to speak about human dignity and autonomy before incorporating these systems into its processes, is a modest signal, but one consistent with the underlying thesis we hold at Zendoric: in the short term, military AI poses real dilemmas about responsibility and human control that do not resolve themselves, and the sensible path is neither to ban nor to accelerate without brakes, but to govern with evidence and judgment from the first contact with the technology, in the classroom and not only in international treaties.

The long-term reading also applies here, albeit with nuances: the same perception and automation systems Navon helped develop for civilian vehicles —object detection, collision avoidance— are the technical foundation both of defensive and rescue applications and of weapons systems. Which of them prevails will depend less on the capability of the model than on the institutions that govern it. That, at bottom, is the message this talk at ACAPOMIL leaves for the rest of the sector: the ethical conversation cannot arrive after deployment.