Where OpenAI’s technology could show up in Iran

OpenAI's recent and controversial agreement with the Pentagon opens the door for its AI models to be used in classified military operations, potentially including the escalating conflict with Iran. While CEO Sam Altman claims the deal prohibits building autonomous weapons and domestic surveillance, critics argue the military's own permissive guidelines offer little real constraint. The company's motivations are unclear, ranging from a need for revenue to fund expensive AI training to Altman's stated belief that liberal democracies must have advanced AI to compete with China. The speed of this pivot from a company once wary of military applications is notable, following a similar path as Elon Musk's xAI, which also struck a deal for its Grok model.

If integrated in time, OpenAI's technology could be used by human analysts to process lists of potential targets, analyzing text, image, and video intelligence to prioritize strikes based on logistics and other data. This would add a conversational layer on top of existing systems like Project Maven. Separately, OpenAI's partnership with defense contractor Anduril focuses on counter-drone technology, using AI for time-sensitive analysis to help take down attacking drones—a application OpenAI argues doesn't violate its 'harm' policies as it targets machines, not people. These developments represent a new frontier: moving from AI for data analysis to using generative AI for actionable combat recommendations, testing the limits of what its customers and employees will tolerate.