Post

P(aul) Frazee
@pfrazee.com

Legitimately we’re ambling toward a world where the only companies able to defend themselves are the ones that are granted access to the AI security programs of 2 or 3 companies, if we ban open weights

P(aul) Frazee@pfrazee.com

Our invitation to glasswing must have gotten lost in the mail

38 reposts325 likes

Time Spinner
@spinny.bsky.social

exclusive 'we' here because there's no chance Mistral, Naver, Baidu, Deepseek, Sakana, Alibaba or Tencent will hit the brakes for the convenience of the American stock market

Finfet
@finfet.sh

the worse part of this is how it impacts downstream companies. it solidifies that you must use products from a company who is blessed with access if you don't want to be put into an insecure position. there will be no alternative to palo alto, cisco, or other existing infra incumbents.

Finfet
@finfet.sh

but hey one way to get access to just let openai or ant accidentally hack you. worked out well for huggingface.

Kevin Riggle
@kevinr.free-dissociation.com

We’re absolutely not going to ban open weight models, because how would that work, but even if we did cybersecurity done by actual human intelligences will continue to truck along fine. The problem as ever is about addressing whole vulnerability classes

max
@notplants.net

i guess that as things adapt to this new climate, we could also get more tools & frameworks that make it harder to mess up security when you use them