Grok used in Iran strikes

In a sworn court declaration by Cameron Stanley, the US Department of War Chief Digital and AI Officer, he claimed that xAI's Grok for Government model was used during Operation Epic Fury to strike 2,000 distinct targets in Iran within 96 hours. This declaration was filed to keep a data centre running in the NAACP Clean Air Act suit against xAI, which is alleging that xAI is using polluting gas turbines in Memphis, Tennessee.

Stanley's statement has been used to justify the data centre as a matter of national security, saying that Grok has allowed the "support [of] military planning workflows, report synthesis and generation, predictive analytics for logistics and sustainment, red-teaming analysis of adversary positioning, personnel management, and medical supply lines".

What's interesting is that in his statement Stanley referred to this use of AI in the war on Iran as something that has "enabled" the Operation, and not that it was used to fire on the targets. He seems to put Grok as assisting in decision-support and back office work, rather than fire control.

At the same time, we have to note that this did not come from a defence briefing but in a filing from a stakeholder interested in keeping a data centre open, so there is of course some bias. Either way, it is the first time anyone in the administration has confirmed the use of Grok in the bombing of Iran.