The War Our Feeds Chose

In partnership with

My Instagram feed looked like a festival. Dancing. Fireworks. Someone handing out sweets on a street corner. Many videos started the same way: "Don't tell us how to feel. We are happy."

It was 6 am on March 1. Two hours earlier, I'd woken to the news that the United States and Israel had struck Iran and Ayatollah Ali Khamenei was dead. I'd lain there staring at the ceiling, failed to go back to sleep, and opened Instagram expecting fury. Got a party instead.

I texted a friend who has spent much of her life in Oman. "What is even happening? Are people not angry?"

She sent me a LinkedIn post by content creator Cyrus Janssen, who had shared a comment from an anonymous Iranian viewer on his YouTube channel. The man described being trapped between two fears at once: a government that had "alienated multiple generations" through decades of suppression, and the terror of what comes after, having watched Iraq, Libya, Syria and Afghanistan get "liberated" into rubble.

"We don't trust the U.S. or Israel. Not because we support our regime — but because we know how imperial powers treat 'liberated' nations in the Middle East." 

He ended with a line that stayed with me.

This was the Iran that my feed wasn't showing me.

What my feed was showing me was the diaspora. Iranians who had spent decades outside the country flooded social media before journalists inside could file their first dispatches — in English, emotionally legible, camera-ready. 

Inside Iran, the picture was messier. Some did celebrate openly — cars honking, people screaming from rooftops in parts of Tehran, Karaj, Isfahan. But people who had been out on the streets during the January crackdowns largely stayed home, as did many who simply didn't know what came next. 

Meanwhile, Al Jazeera showed me thousands pouring into Tehran's streets in mourning, with 40 days of official grief declared by the state.

The algorithm had shown me one slice of a country holding three contradictory truths simultaneously. Diaspora celebration in English moves faster on global social media than frightened ambivalence in Farsi. The feed is never neutral, and I was reminded that once again. 

The part that reads like a spy thriller. With caveats.

Once the dust settled, the internet's next obsession was: how did they find him? Stories multiplied, suggesting prayer app hacks, dental implant trackers and surveillance cameras.

I want to be honest about what we know and don't. Most intelligence reporting on the operation comes from unnamed officials speaking to outlets like the Financial Times. In a war this opaque, distinguishing intelligence briefing from strategic narrative is genuinely difficult.

That said, according to several accounts, Israeli intelligence had compromised Tehran's traffic camera network long before the strike, using cameras near Khamenei's compound to map the routines of his security team. The CIA had been tracking Khamenei's location and patterns for months, and the strikes were planned around a confirmed meeting of senior officials.

The irony isn’t hard to miss. Iran built its surveillance infrastructure to monitor its own protesters and dissidents. If these accounts are accurate, the same cameras were simply turned around. The surveillance state, surveilled by a different surveillance state. And if that's possible in Tehran, it's worth asking: who is really watching the cameras watching you?

Meanwhile, open-source researchers were doing something the spy-thriller threads weren't: reconstructing civilian casualties. OSINT investigators geolocated footage from Minab showing a US Tomahawk cruise missile hitting a military compound next to a girls' elementary school. Over 150 girls between seven and twelve were killed at morning classes. 

And then there's the story nobody was telling on Instagram.

The day before the bombs fell, something extraordinary happened in Silicon Valley. The Pentagon had given Anthropic, the American AI company behind Claude, a deadline: remove restrictions on mass domestic surveillance and autonomous weapons, or lose the contract. Anthropic refused. The Trump administration then designated it a "supply chain risk to national security", a label previously reserved for Chinese companies like Huawei.

Hours after the bombing began, OpenAI announced its own Pentagon deal. Sam Altman later admitted the timing "looked opportunistic and sloppy." 

Users ditched ChatGPT and pushed Claude to the top of the App Store — a strange kind of wartime reward for a company just declared a national security threat. OpenAI's top robotics executive resigned over the deal, saying that "surveillance of Americans without judicial oversight and lethal autonomy without human authorisation" deserved more deliberation than they got.

These models process satellite imagery, intercepted communications, and population-scale metadata at speeds no human analyst can match. The Iran operation may be the first time that capability was deployed in a live conflict at this scale. The companies building the tools and the governments using them are now in a negotiation the rest of us aren't part of — and whose terms keep changing mid-war.

Meanwhile, on Indian television, my colleague Hera Rizwan was documenting a completely different war. Headlines read: "In a few hours Iran's skies will be under US control." Channels aired old footage from Bahrain and presented it as Dubai under attack. This wasn’t just sloppy journalism. Tens of thousands of Indian students study in Iran every year, and millions more have relatives across the Gulf. For parents already scrambling to reach their children, these broadcasts didn’t just misinform — they amplified fear. In many homes, television became the only window into a crisis they had no other way to verify.

Read Hera’s full piece here.

So, Instagram showed diaspora celebration. Al Jazeera showed mourning. OSINT investigators found a school in Minab. Indian television showed apocalypse. Silicon Valley fought over who gets to build the next targeting system. Five versions of the same morning.

Most of us never see the physical battlefield or the intelligence infrastructure behind it. We experience the information one. That battlefield is shaped by algorithms, government narratives, media chasing ratings, and AI companies negotiating their ethics in real time — sometimes in the middle of an active war.

Which leaves the question I had at 6 am, still unanswered.

Who do we believe?

The celebration was real. The mourning was real. The man stuck in a house he hates, surrounded by fires he fears more, he's real. The school in Minab is real. The AI contract is real. None of them are the full story. The actual Iran — 90 million people holding three contradictory truths at once — cannot fit in an Instagram reel, a primetime headline, or a targeting coordinate.

The version of reality that reaches us first is almost never the complete one. That was true before algorithms. It's just faster now.

On My Bookmarks

Got a story to share or something interesting from your social media feed? Drop me a line, and I might highlight it in my next newsletter.

See you in your inbox, every other Wednesday at 12 pm!

AI Agents Are Reading Your Docs. Are You Ready?

Last month, 48% of visitors to documentation sites across Mintlify were AI agents—not humans.

Claude Code, Cursor, and other coding agents are becoming the actual customers reading your docs. And they read everything.

This changes what good documentation means. Humans skim and forgive gaps. Agents methodically check every endpoint, read every guide, and compare you against alternatives with zero fatigue.

Your docs aren't just helping users anymore—they're your product's first interview with the machines deciding whether to recommend you.

That means:
→ Clear schema markup so agents can parse your content
→ Real benchmarks, not marketing fluff
→ Open endpoints agents can actually test
→ Honest comparisons that emphasize strengths without hype

In the agentic world, documentation becomes 10x more important. Companies that make their products machine-understandable will win distribution through AI.

Make Your Docs Agent-Ready

Was this email forwarded to you? Subscribe