This series started with what looked like a small glitch: asking an AI to generate images tied to the “Axis of Resistance” somehow made it weirdly stubborn. But honestly, digging into that brought up a deeper issue: what kinds of politics are hiding inside artificial intelligence itself? Now, in Part II, we’re leaving the screen behind and going underground to the systems and deals that make all this possible in the first place. Because the real story isn’t just what some chatbot spits out. It’s about where the data sits, who holds the keys to those massive server farms, which companies rent their tech out to the military, who’s building the surveillance tools, and which governments can turn ordinary tech giants into silent partners of state security.
Let’s cut to Israel’s Unit 8200; it’s the clearest example of this mess. So, take Unit 8200 and the cloud. Back in August 2025, The Guardian, +972 Magazine, and Local Call broke a wild story: Unit 8200 built a gargantuan surveillance system, and they did it using Microsoft’s Azure cloud. That means millions of phone calls from Palestinians in Gaza and the West Bank ended up warehoused in Microsoft’s servers. "A million calls an hour," according to insiders. The system fired up in 2022, supposedly after Microsoft CEO Satya Nadella met with Unit 8200’s commander, Yossi Sariel, the year before.
Just think about that for a minute. When people talk about “the cloud,” you picture something airy—servers, Google Docs, AI models floating in cyberspace. But here, “the cloud” was this gigantic file cabinet, jam-packed with intercepted calls from a whole occupied population. Microsoft’s Azure gave Unit 8200 the muscle: the space to store all this data, and the horsepower to sift through it way beyond what a traditional army IT department could pull off. And that’s the shift: Big Tech isn’t just selling gear to the army. Now, its infrastructure can become central to how intelligence is gathered and used. Once that data lands in the cloud, it doesn’t just sit there. AI, and especially machine learning, starts looking for patterns.
What’s the value in all this eavesdropping? It’s not just hearing someone talk, it’s connecting dots a person couldn’t. Who calls whom? Where do these phones go? Which numbers keep popping up together? Who shifts their habits before something goes down? The relationships and connections the stuff that, on this scale, only algorithms can really sort out. That’s why the dogpile of Unit 8200, big cloud providers, and AI deserves a closer look. Microsoft wasn’t just warehousing calls; they gave the unit a way to turn basic surveilling into full-on, industrial-scale analysis. Per the investigation, Unit 8200 used this pile of intercepted data to pick out targets for bombing runs in Gaza. That changes the whole equation. Technology isn’t off to the side of war zones anymore, it’s not “back office.” The cloud is now part of the battlefield.
Here’s where things twist a bit: Microsoft drew a line. Eventually, they took action. After the exposé, Microsoft launched an inquiry. They said using Azure for mass civilian surveillance was against their rules. On September 25, 2025, they cut off Unit 8200’s access to the relevant storage and AI services. Why is that a big deal? First, it proves tech companies can set limits for the military if they want to. Second, it shows why journalism, activism, and public pressure matter. Microsoft workers and former staff protested; there were even demonstrations. This system didn’t fix itself. It took reporting, dissent, and outside pressure to force action. But even then, Microsoft didn’t sever every military tie with Israel. That line matters.
Now, look at Project Nimbus - Google and Amazon’s approach. In 2021, those two snagged a $1.2 billion Israeli cloud contract. The reporting showed wild contract clauses, ones that protected Israel’s access to the cloud, even if Google or Amazon wanted to pull out. For instance, if data was ever handed to foreign authorities, Israel got a heads-up. Plus, the companies contractually limited their own ability to pull services because of policy changes or ethical concerns. That’s not just shopping for cloud services; it’s bargaining for the political terms under which those services might be yanked away. And it says a lot about how crucial cloud infrastructure is, strategically. Israel saw future political pressure coming from Google’s and Amazon’s workers, shareholders, courts, governments—everywhere. So they locked in protections to make sure the cloud couldn’t just disappear if a company got cold feet. Compare that to Microsoft’s later move. Microsoft ended the deal based on their rules:
"This breaks our terms; you’re out."
But Project Nimbus was designed to make that kind of unilateral cut-off much harder to pull off. This blurring of lines (the “military-cloud complex) means we have to drop the fantasy that Silicon Valley and the military are two totally separate worlds. Far from it. Cloud companies store the data. AI companies build the models. Cybersecurity outfits invent hacks and shields. Satellite companies beam communications. Social media collects everyone’s lives. Chip makers fuel the computing power. Intelligence agencies toss in oceans of raw info. Put all that together, and you get the next-gen military-industrial complex. Israel plays a starring role because of its tight web of military intelligence, tech startups, universities, VCs, and global tech giants. Unit 8200 acts as a conveyor belt—personnel come out, found startups, build security tools, attract investment, get bought out by multinationals—and the military’s innovations spread worldwide.
Pegasus showed the sinister side: Israeli spyware turns your phone into a surveillance device. Unit 8200 shows the scale: not just targeting a person, but hoovering up and processing millions. Cloud turns the dial even higher, surveillance goes industrial. And AI? That automates interpretation itself. This brings us right back to AI censorship. If the firms running AI infrastructure are so deeply bound up with U.S. security agencies, Israeli companies, the Israeli military, and Western governments, is it realistic to think their AI systems are neutral, politics-free? Nope. But let’s not be dramatic: Israel’s not telling every model what to say. The deeper issue is how structure works. Power shapes the tools; tools shape data; data shapes the algorithms; algorithms affect what you believe. And what you believe? That shapes politics.
Ask an AI about Hamas, Hezbollah, the IRGC, Israel, Palestine, or the Axis of Resistance, and you’re not chatting in some sterile void. You’re getting responses shaped by layers of influence - government, corporate policies, legal frameworks, journalists, and the tech sector. Some groups get filed as “governments,” others as “terrorists.” Some attacks become “self-defense.” Others are “terrorism.” Some intelligence is reported as fact. Other bits are dismissed as propaganda, unless someone can verify it. Those labels matter. And as the world’s knowledge infrastructure gets more tangled with the agendas of powerful states, “neutrality” isn’t something you can just assume. It has to be proven, over and over. The cloud is now a battleground. And that sets up a tougher question for Part III: If Big Tech supplies the infrastructure for surveillance and warfare, what happens when those same companies control the AI that most people use to understand, or even see what’s going on in those wars?
We’re moving from technology serving power, to technology shaping how people think about power itself. That’s a whole different territory.
No comments:
Post a Comment