Post

Shakeel
Shakeel@ShakeelHashim·
I suspect that if we could figure out a way to make open weight models without dangerous capabilities — eg tamper resistant architectures or machine unlearning — Anthropic would be in favor of doing so (so would I!).
Sara Hooker@sarahookr

Between the lines, Anthropic is saying “I’m okay with open weights as long as the model does not present high performance.” “Release what you want as long as not a direct alternative to proprietary models.”

English
3
0
62
48.8K
Joshua Saxe
Joshua Saxe@joshua_saxe·
@ShakeelHashim Cyber capabilities are all dual use though, so in this setup defenders have to pay closed providers to protect themselves with pentesting and vuln finding and there's no startup and oss ecosystem around hacking and fine tuning open models
English
2
0
6
284
David Manheim
David Manheim@davidmanheim·
@ShakeelHashim Sure, but I suspect that everyone is putting too little weight on the fact that BOTH SIDES of the argument are mostly driven by companies fighting over shares of the profits from AI, not philosophy, or security, or values. @sarahookr x.com/davidmanheim/s…
David Manheim@davidmanheim

Let's be clear - this isn't about AI security, it's about the market power that open-source AI gives NVIDIA and the direct economic benefits to many of the other signatories to the pro-open-weights cartel.

English
1
0
2
297
Paylaş