> The idea that AI companies scaremonger to sell models is a silly meme.

> Both OpenAI and Anthropic were founded by people who sincerely believed in the risk of out-of-control superintelligence.

This is not a contradiction.

These things can all be true:

1. That they were afraid of ASI

2. That they continue to be afraid of ASI

3. That they recognize that LLMs aren't in fact a path to ASI

4. That the current models aren't the existential danger they'd have us believe

5. That they're claiming they are because it makes for good marketing