| ▲ | bottlepalm 2 hours ago |
| There are thousands of data centers around the world with machines capable of running these large models. You don’t have the access or jurisdiction to turn them all off. |
|
| ▲ | stephantul 2 hours ago | parent | next [-] |
| Ok but do any of these data centers have a copy of the models that attacked hf? |
| |
| ▲ | stymaar 2 hours ago | parent | next [-] | | I think that the argument is that an hostile model could attack overseas datacenters, host itself there and then launch its attack from there. | | | |
| ▲ | Arainach 2 hours ago | parent | prev [-] | | Given that the models have been proactively hacking other companies, why does it matter where the code currently is? It could move to any of them. | | |
| ▲ | stephantul 2 hours ago | parent [-] | | Not the code: the weights. Are you going to host a trillion parameter model somewhere without someone noticing? | | |
| ▲ | dumberquestions 2 hours ago | parent [-] | | It's not unthinkable, do you think all cloud providers with sufficient compute have perfect monitoring? | | |
| ▲ | ThrowawayR2 an hour ago | parent [-] | | Cloud providers want to make sure they bill someone for every microsecond of compute, particularly GPU compute which is in demand. They constantly work towards good monitoring because their profits are directly on the line. | | |
| ▲ | bottlepalm 28 minutes ago | parent [-] | | I’m sure a super AI can find/scam enough money to pay its own bills under the cover of doing something legitimate. |
|
|
|
|
|
|
| ▲ | PowerElectronix 2 hours ago | parent | prev [-] |
| The owners would very much turn it off as soon as they see workloads freeloading in their machines. Unless you suggest the LLM would foot the bill somehow. |
