Anthropic has put itself in a tough spot

Anthropic invited Christian church leaders for a two-day retreat on the moral compass of AI — fitting the "responsible" brand it cultivates. But as I told NOS, the company is in a strange split: it positions itself as the safety-first AI lab while simultaneously being the biggest accelerator, constantly pushing out new models. They keep egging each other on — "if the real bad guys are going to do it, we'll at least do it the right way."

The result is a company that signs a $200 million Pentagon deal but draws red lines on autonomous weapons, builds a model it calls too dangerous to release publicly, and sits down with clergy to discuss whether AI could be divine. Let's not praise these people to the heavens for their morality, as TU Delft ethics professor Jeroen van den Hoven put it — but by taking these positions publicly, they've at least committed themselves to standards they can be held to. OpenAI, by contrast, goes whichever way the wind blows.

Link to original post here