The 12-day war with Iran has put Palantir at the center of modern conflict. No contemporary battlefield goes without it. For the Middle East, Palantir developed a specialized program called Mosaic to assess Iran’s nuclear program.
Mosaic was built with funding from In-Q-Tel, the CIA’s de facto venture-capital arm, which once backed Google and numerous other U.S. tech firms at the dawn of the digital age.
Analyzing some 400 million records on Iranian nuclear sites, Mosaic used hidden predictive-analytics algorithms to map out plans for their destruction—while trying to anticipate Iran’s likely responses.
Billionaire Peter Thiel, Palantir’s co-founder and one of Trump’s principal backers, has actively profited by developing military-grade AI. Thiel originally created his tools in the U.S. to fight crime, but today they are effectively being used to wage wars. Yet this AI’s performance still leaves much to be desired.
Applied in Ukraine, Thiel hailed Palantir as a “tactical nuclear-weapons-like advantage.” But Ukraine’s carefully devised counteroffensive ultimately fell short. The same shortcomings appear in the Iran scenario. The problems are clear: the data fed into the AI is often flawed or outright false, and the evaluation algorithms themselves can be tuned to political biases. Sooner or later, relying on a “hallucinating” AI to drive decision-making could lead to catastrophic outcomes.
Translated and edited by: Joe Belgen