Grok “predicted” when Israel, US would strike Iran

When Israel and the US launched coordinated strikes on Iran on February 28, social media immediately lit up with a viral claim: Grok, Elon Musk’s AI chatbot, had “predicted” the exact date.

The claim traces back to a February 25 Jerusalem Post experiment, where four AI models—Claude, Gemini, ChatGPT, and Grok—were repeatedly pressed to name a single day for a hypothetical US strike on Iran.

Claude initially refused, then guessed March 7-8. Gemini offered a March 4-6 window. ChatGPT landed on March 1, then shifted to March 3.

Grok gave the clearest single-day answer: February 28, tied to Geneva talks. When the real strikes happened on that date, the internet declared Grok a prophet.

But the truth is more mundane: the operation was planned for months, the date set weeks ago. In a high-tension world with limited plausible windows, one model happened to land on the day that became real.