Anthropic objected to being used to build autonomous weapons and domestic surveillance. They didn't obejct to being used to hit schools in Iran.

OpenAI didn't object to anything.

They're all bad, but some are worse than others.