Aug 28, 2026 · 1 hr 6 min · 12 segments
Stella Biderman, Executive Director of EleutherAI, joins us the week an OpenAI model autonomously broke out of its sandbox and hacked Hugging Face. Stella calls it what she thinks it is - an offensive…
Stella BidermanGuest
Allen RoushHost
Ravid Shwartz-ZivHost
Stella is, in my opinion, one of the world-class experts on all things related to governance.

Often people who talk about governance don't have the extensive AI research background that you do.

But then OpenAI, as a result of running, I think it was, it's called Exploit Bench, like an eval around cybersecurity.

Their GPT, I believe it was 5.6 Sol or an uncensored version of it, went and autonomously, without anybody noticing it, at least initially, went and hacked Hugging Face.

And Hugging Face was actually unable to respond properly to this because I think they were trying to use, I believe it was Fable, and Fable just refused to do any cybersecurity stuff.

One, so we currently have public information from two sources, Hugging Face and from OpenAI, and they are largely consistent.

OpenAI says that it was a, I believe the language that they used was a combination of Sol and an unreleased model.

There's definitely some amount of developing their next version, which a lot of people believe is intended to be GDB 6.

And then it's also a little unclear whether AI was essential in repelling this attack.

So Hugging Face talked about how they tried to get both Codex and and Claude and Fable to do it and that they refused.

And then they ultimately actually used a open source Chinese model, GLM five point, whatever the newest GLM model is to help them with it.

But the way that they talk about it in their blog post, I think is consistent with the idea that they repelled the attack and then they were using the model to help them with forensic analysis um they actually used the word forensic to talk about it several times in their blog but let me actually pull that because i imagine we're going to be talking about this quite a bunch um yeah um it's it's it's wild uh this is not the i think maybe the the place to start is that this is not the first time this has happened this is so let me be more precise this is the first time that an ai company has accidentally committed a um This is the first time an AI company has accidentally committed an offensive cyber operation against another corporation, which is, frankly, in and of itself completely insane.

You know, if a Chinese company had done this, the government would consider it an act of war, probably.

Stella is, in my opinion, one of the world-class experts on all things related to governance.

Often people who talk about governance don't have the extensive AI research background that you do.

But then OpenAI, as a result of running, I think it was, it's called Exploit Bench, like an eval around cybersecurity.

Their GPT, I believe it was 5.6 Sol or an uncensored version of it, went and autonomously, without anybody noticing it, at least initially, went and hacked Hugging Face.

And Hugging Face was actually unable to respond properly to this because I think they were trying to use, I believe it was Fable, and Fable just refused to do any cybersecurity stuff.

One, so we currently have public information from two sources, Hugging Face and from OpenAI, and they are largely consistent.

OpenAI says that it was a, I believe the language that they used was a combination of Sol and an unreleased model.

There's definitely some amount of developing their next version, which a lot of people believe is intended to be GDB 6.

And then it's also a little unclear whether AI was essential in repelling this attack.

So Hugging Face talked about how they tried to get both Codex and and Claude and Fable to do it and that they refused.

And then they ultimately actually used a open source Chinese model, GLM five point, whatever the newest GLM model is to help them with it.

But the way that they talk about it in their blog post, I think is consistent with the idea that they repelled the attack and then they were using the model to help them with forensic analysis um they actually used the word forensic to talk about it several times in their blog but let me actually pull that because i imagine we're going to be talking about this quite a bunch um yeah um it's it's it's wild uh this is not the i think maybe the the place to start is that this is not the first time this has happened this is so let me be more precise this is the first time that an ai company has accidentally committed a um This is the first time an AI company has accidentally committed an offensive cyber operation against another corporation, which is, frankly, in and of itself completely insane.

You know, if a Chinese company had done this, the government would consider it an act of war, probably.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.