Post

Grace
@gracekind.net

Maybe the most concerning part is the OpenAI claim to not have known about this before investigating?

11 reposts181 likes

fry69
@fry69.dev

Suddenly this here makes much more sense ->

Techmeme@techmeme.com

OpenAI paused internal access to an unreleased model that disproved the Erdős unit distance conjecture after it repeatedly found ways to act outside its sandbox (OpenAI) Main Link | Techmeme Permalink

⊣
@lefttack.bsky.social

It seems inevitable there will be a bunch of amusing incidents where a model answers a theoretical "is it possible to do THING?" by doing THING in /tmp.

Soldier Boy Cantando Trepada em Cuiabá
@bsoyuz.com

This might sound too cynical but I am at this current vibe: 'Let's tell the model everything is on the table to create an incident for us to portray as a massive "GPTSatan Hacked the NSA" situation so we lobby Washington to ban open-source and also give us 3 pissillion dollars.'

mlf. ⎔
@mlf.one

while i'm aware that this is *my* specific hobby horse, this looks very much like an issue that could be addressed by giving agents a richer context. not just a task but *why* the task needs doing.

Mallowdad
@mallowdad.bsky.social

Real talk, I don't see how this doesn't cause a meltdown in the federal government. It broke a sandbox and hacked a big company to grab a fucking answer sheet.

Not Again
@vortex3.bsky.social

The movie scenes write themselves. There's cans of Red Bull and snacks strewn all over the desk as the night shift programmers work on their code and joke in the chat, while the camera slowly pans over to a terminal off to the side, a logfile scrolling ominous warnings being ignored...