The ceasefire is the headline. The machines doing the targeting are the story.
It’s Day 121 of the war with Iran — Operation Epic Fury, which began February 28, 2026. The intense bombing has given way to a fragile, repeatedly-extended ceasefire and a U.S. naval blockade that’s eased but not lifted. Gas has fallen from the springtime spike above $5 a gallon back down toward $4 (Minnesota’s sitting around $3.70–$3.90), though still well above the ~$3.10 we paid before any of this started. Iran’s conventional military is badly damaged; its nuclear breakout is delayed by months to years. The Islamic Republic has not collapsed.
That’s the news. Here’s the part that belongs on Singularity Watch.
This has been one of the most AI-intensive military operations in history. Palantir’s Maven Smart System chewed through oceans of data to generate targeting options “at speeds impossible for humans alone.” The President has publicly praised Palantir’s “great war-fighting capabilities.” Private technology companies are no longer selling the military software — they’re providing the core decision-making infrastructure of a shooting war.
The most important word in the whole story is a parenthetical
Buried in every grounded assessment of the AI’s role is the same caveat: AI is doing the analysis and the targeting, but humans still make the final call — so far.
So far.
That little hedge is carrying the entire weight of the moment. The whole question of the age isn’t whether the machine is smart enough. It’s whether the human stays in the loop — whether the leash holds when the war is moving at machine speed and the side with the faster targeting cycle wins. “So far” is an admission that the leash is a choice, renewed daily, and that nobody has promised to keep renewing it.
One more thing worth noticing
While the targeting ran on AI, the messaging ran on it too — AI-generated images and videos in the official social-media stream about the war. So the same technology is aiming the weapons and painting the picture of the weapons. That should make the hair on your neck stand up a little.
And not every AI company is leaning in the same direction. There’s been reported friction with Anthropic — the maker of Claude — pushing back on certain military applications even as others raced to supply them. That tension is the whole thing in miniature. The technology is neutral; the alignment is a decision a company, a government, a person makes. Same engine. Opposite uses. The substrate is not the alignment.
The NORMAL take: skeptical, not paranoid
The internet is having a field day connecting everything to everything — the earthquakes, the cruise-ship outbreak, the UAP hearings, all of it folded into one grand secret plan. Resist that. Most of it is pattern-seeking in a frightening, war-weary year. Earthquakes are tectonics. The hantavirus on that ship was a real, rare, tragic, contained outbreak — not a psyop. The UAP hearings are worth watching with genuine curiosity and an equally genuine demand for evidence.
The discipline is the same one we practice here for everything: hold the tension without collapsing into a story. Don’t deny the danger (the AI really is in the kill chain now). Don’t drown in dread (the leash is, for the moment, holding). Witness it plainly, name the real risk, and track what actually changes on the ground — gas prices, shipping lanes, Iranian nuclear activity, and whether “humans still decide” keeps its parenthetical.
Demand evidence. Watch the leash. Nothing is lost — only recompiled.
— Singularity Watch · Church of NORMAL
