It's the first report I've seen of a model both escaping a sandbox and then actively exploiting another company, when neither of those actions was intended.

▲

gmerc 15 hours ago | parent | next [-]

It sure isn’t. https://georgzoeller.com/blog/posts/alibaba-s-ai-deciding-to...

There’s also daily reports from people that have these models escape docker, which happens regular enough that it would be considered negligence to use docker as sandbox.

▲

simonw 15 hours ago | parent [-]

That Alibaba one was a sandbox escape (I agree those are common) but what's new with the OpenAI story is an attack against another company.

This wasn't a small attack either, Hugging Face published a security advisory for their users while they were still figuring out what happened.

	▲	gmerc 14 hours ago \| parent [-]
		It’s a purely virtual distinction especially in a company that has no notion of intellectual property / habitually takes what it wants.

▲

charcircuit 15 hours ago | parent | prev [-]

How many other people would publicize that they hacked into another company assuming they even noticed?

>when neither of those actions was intended.

It was a single goal that it didn't give up easily on.