- Autonomy can cause catastrophic errors when speed outweighs human judgment
- The Minab case shows how AI can misread data with deadly consequences
- As telecom deploys more and more automation, safeguards and accountability are essential
The telecommunications industry has spent many years in pursuit of the highest levels of autonomous networking possible. From the keynote stages of MWC to the boardroom suites of Tier-1 carriers, the gospel of Autonomous Networks has been preached with religious fervor.
The journey has been long and arduous, and for telecom carriers, it is far from over. Many still operate networks at Level 2 or 3 of the TM Forum stack. Only a handful have attained the 4th level of autonomous enlightenment.
What everyone wants to know, of course, is what lies beyond that frontier. When will Level 5 implementations—OpEx’s Holy Grail, fully autonomous and operating without any human intervention—arrive? How will they perform? And can’t we hurry the whole thing up a bit?
This year we got the first documented case study of what happens when organizations push the autonomous envelope too hard, too quickly. But it didn’t come from an R&D laboratory. It came from the sky over a town called Minab.
On February 28, 2026, during the opening hours of Operation Epic Fury, America’s digital kill chain came closer than ever to the destination the telecom industry is navigating toward: full automation.
Amid an opening salvo of nearly 900 missile strikes, one—and one is all it takes—flew with haunting precision. Guided by seven redundant navigation systems and a high-definition live feed, a 3,000-pound Tomahawk missile struck the Shajareh Tayyebeh primary school in Minab.
The technology functioned exactly as programmed, killing an estimated 156 people, including 120 children between the ages of 7 and 12.
America’s official report into the massacre focused on the what, not the how. But subsequent independent investigations confirm the target was chosen using Project Maven—the algorithmic engine perfected by Palantir after Google withdrew in 2018 following thousands of employee protests.
Independent satellite analysis from organizations including Human Rights Watch shows the school was physically partitioned from the neighboring military base as early as 2016, which should have removed it from the target list automatically.
What went wrong? And how can the telecom industry learn from this catastrophe?
Maven’s promise was speed: a roughly $1 billion targeting system that could allow a 20-person cell to perform work that once required about 2,000 staff—a 99 percent reduction—and push toward 1,000 targeting decisions per hour.
But at that tempo, “human in the loop” becomes a legal fiction. Even the more generous public math gives each targeteer barely over a minute per decision; some accounts put the review window at about 48 seconds. That is not enough time to verify source data, challenge an AI-ranked target, assess civilian status, understand the interface output and make a lawful judgment about force.
Department of Defense Directive 3000.09 requires autonomous and semi-autonomous systems to preserve “appropriate levels of human judgment,” undergo rigorous testing and verification, use understandable interfaces, and provide clear activation and deactivation procedures. A process moving at machine speed may satisfy the paperwork of human approval while hollowing out its substance—functionally pushing the kill chain toward Level 5 autonomy, where the machine frames the choices and the human merely clicks.
This represents a chilling epistemic blind spot: the system was programmed to prioritize network connectivity (the digital handshake of the target node) over visual confirmation (the colorful murals on the playground wall, visible through the video camera in the nose cone of the 3,000-pound missile).
For CTOs rushing to integrate AI into their network operating systems, Minab is a chilling portent. It represents the ultimate failure of algorithmic certainty over physical context.
To the AI, the children were merely biological noise on a verified node. The system was tuned according to the same rules as telecom: maximum speed, maximum reliability. And in a network, every packet must be delivered with 99.999% certainty.
This pell-mell race toward autonomy is being fueled by a dangerous regulatory vacuum in the U.S. While the EU scrambles to codify AI ethics, the U.S. remains a Wild West of unregulated corporate R&D. Without a federal framework to govern AI-driven infrastructure, we are essentially beta-testing lethal logic in real time.
In war, the melding of Big Tech AI data-scraping with mil-tech kinetic delivery has become an almost unstoppable force, creating a feedback loop where speed is the only metric that matters. In this narrative, as in Iran, the human inevitably loses their place in the hierarchy of killing. Speed is gained. Responsibility is delegated—to battlefield automata: mindless, amoral, and beyond accountability. This is a trajectory as inevitable as the arc of an artillery shell.
Militaries should not be permitted to use it. It is an abomination—anonymizing and expediting slaughter, disassociating the killers from the killed.
In telecom, the risks might seem less dramatic than on a battlefield. A misrouted packet is not a missile strike. But as networks become more central to critical infrastructure—energy grids, transport systems, emergency services—the consequences of failure grow. A fully autonomous system that misclassifies a network anomaly could cascade into outages, disruptions or worse.
Let us, as an industry, not march blindly into the dark night of Level 5 automation.
We must take a stark lesson from this catastrophe: if telecom continues to accelerate communications networks by removing human oversight in pursuit of Zero-Touch efficiency, we aren’t just building faster networks—we are building the next generation of autonomous failures.
And when Level Five goes wrong, the collateral damage is always human.
Sources:
Stephen M. Saunders MBE is a communications analyst and USPTO-registered inventor examining how digital infrastructure — 5G, cloud, and AI — is reshaping industry, power and society, as well as underpinning the emerging, ubiquitous global digital economy. As anchor of FNTV and a longtime industry insider, he focuses less on growth narratives and more on execution, risk and how hyperscale technology is distorting markets, governance and society at scale.
Opinion pieces from industry experts, analysts or our editorial staff do not necessarily represent the opinions of Fierce Network.