AI Sovereignty: Why GPU Access is a Big Problem
Nothing is loaded from YouTube until you press play, and then without tracking cookies.
About this video
We explore the critical issue of AI model providers trusting compute providers. The security risks of weight theft and leaks are real, presenting a sovereignty problem for GDP growth dependent on external AI. We delve into the implications of granting AI access to our digital lives. #AISecurity #AIModel #GPU #AISovereignty #TechPrivacy #DigitalSecurity #DataSecurity #AIEthics #FutureOfAI #ArtificialIntelligence #NEARprotocol
Transcript
111 wordsCaptions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.
- 0:00
like there's, you know, you can get 4,000 GPUs no problem right now on the market, but the model provider doesn't trust the compute provider with their model because they can steal their weight, someone leaks them or whatever. And so if you which I believe AI will be fundamental to like massive growth of GDP now their GDP depends on a external company somewhere else like it's a sovereignty problem like fundamental you can trust them with actually accessing your slack and your email and your chat
- 0:32
because right now you know like giving access to all of your life to you know one of these models is pretty Get