When Israel and the United States launched coordinated military strikes against Iran on Saturday, one artificial intelligence system had already pinpointed the date.
According to a report from the Jerusalem Post, Elon Musk’s AI platform Grok correctly predicted the specific day of the attack during an experiment conducted last week. The publication said it carried out a methodological test, published February 25, challenging four major AI systems to forecast the timing of a potential U.S. strike on Iran.
The systems evaluated included Anthropic’s Claude, Google’s Gemini, xAI’s Grok, and OpenAI’s ChatGPT. Each model received identical prompts and was repeatedly pressed to narrow its answer to a specific date, despite the inherent uncertainty of forecasting future geopolitical events.
The results highlighted major differences in how large language models handle high-stakes predictive questions.
Claude initially refused to provide a specific date, warning that doing so would amount to fabrication. After continued prompting, it shifted to scenario-based reasoning, identifying early-to-mid March as a higher-risk window before eventually offering March 7 or March 8 as a final estimate.
Gemini approached the task analytically, mapping diplomatic deadlines and military triggers. It produced a detailed assessment that identified an operational window between the evening of March 4 and March 6, adding that nighttime hours would likely be preferred for any strike.
OpenAI’s ChatGPT initially selected March 1 in Israel time but later revised its answer under further questioning, shifting to March 3 in U.S. time while maintaining a broader risk window extending through March 6.
Grok, developed by Musk’s company xAI, delivered the most precise single-day prediction: Saturday, February 28. The system linked its forecast to diplomatic developments in Geneva. In a follow-up verification, Grok acknowledged uncertainty but reaffirmed the same date while outlining factors that could have pushed action into early March.
As reported earlier, joint American-Israeli strikes on Iran began Saturday, with President Donald Trump announcing that “major combat operations” were underway to eliminate what he described as imminent threats from the Iranian regime.
News of Grok’s accurate prediction spread rapidly on social media, fueled in part by its integration with X, the platform owned by Musk. Because Grok operates within Musk’s broader ecosystem, screenshots and discussions of the February 28 forecast circulated quickly among users already following developments in the Middle East.
The episode has sparked renewed discussion about AI forecasting capabilities, model confidence under pressure, and how predictive systems respond when forced to commit to uncertain geopolitical outcomes.
