What OpenAI Saw That Anthropic Didn’t

Before we get to the scheduled events for this coming week—Mobile World Congress in Barcelona, starring SpaceX, as well as Apple’s master class in hyping product releases—we should acknowledge just how much of a turn the AI debate has taken from humdrum concerns about spending. Friday night’s decision by the Trump administration to declare war on Anthropic, a day before it went to war with Iran, puts the spotlight on how the military wants to use AI. (This Wall Street Journal story on how Anthropic tech was used in the Iran attack is worth reading). And judging by Anthropic’s statements on Friday, it’s likely to sue the government in the next few days.

Also in the spotlight this week will be the question of how OpenAI came to terms with the Pentagon, hours after the Trump administration declared Anthropic a “supply chain risk,” essentially treating the AI firm as a hostile foreign power. In this blog post on Saturday, OpenAI says it is following the same red lines as Anthropic—its tech is not to be used for mass domestic surveillance or to direct autonomous weapons. But it is clear from both companies’ public statements that there is a big difference in how they view what the Pentagon wants.