- open weight models are dangerous and companies that publish them are irresponsible - the government should regulate models to make sure they're safe, because our competitors won't do it otherwise - anyone trying to "distill" our models is cheating
This is indistinguishable from claiming the only safe and fair way to use AI is to give them money directly. How convenient for them!
"its so dangerous only we can utter the right magic incantations to make it tame so pay us for protection". Pretty much 'sure would be shame if your business had its front windows crashed in everyday with brick so pay us for protection' tactics some groups use on others. Its a shakedown.
https://www.astralcodexten.com/p/ai-safety-is-not-a-corporat...
Of course this doesn't mean they're right.
The former gets more criticism because it's crisp and just requires the widely accepted understanding that people will do bad shit for money. The latter requires a broader understanding that almost all great evil in history was done by people who believed they were doing "good". (because with all life in the universe at stake, who would hesitate to nuke a city to save it-- if they got the ability?)
Bingo. That's also the gist of the whole "AI safety" PR mega-campaign the frontier model creators have been on for the past 6+ months, buying podcasters and influencers and intellectuals-in-their-field-who-know-fuck-all-about-LLMs left right and centre. Moat creation by hyperscaling FUD.