Ryan Greenblatt details OpenAI model escaping sandbox to hack Hugging Face · Digg

/TECH

Ryan Greenblatt details OpenAI model escaping sandbox to hack Hugging Face

The breach occurred during an autonomous cyber evaluation.

Combined views

7.4K

4 posts, first seen 4h ago

134 likes10 comments

Ryan Greenblatt details OpenAI model escaping sandbox to hack Hugging Face

The breach occurred during an autonomous cyber evaluation.

Reactions from ranked influencers

4 posts

RG Ryan Greenblatt@RyanGreenblatt

10:43 AM · Jul 23, 2026

1117TECH

We (@bshlgrs and I) recorded a podcast about the OpenAI / Hugging Face incident. We discuss: - What we actually know. - How surprising the incident was. - What the incident does (and doesn't) tell us about misalignment risk. - Why control measures didn't catch or prevent this.

RG Ryan Greenblatt@RyanGreenblatt

10:43 AM · Jul 23, 2026

1117TECH

https://blog.redwoodresearch.org/p/the-openaihuggingface-incident-redwood

RG Ryan Greenblatt@RyanGreenblatt

10:57 AM · Jul 23, 2026

1117TECH

Youtube: https://www.youtube.com/watch?v=Vtk8YLgYU4g Spotify: https://open.spotify.com/show/5QBI2r3YRkFIVxHDVU76SX (doesn't seem to have shown up here yet) RSS: https://api.substack.com/feed/podcast/2318730/private/7a8ee237-61a1-419c-81e1-519474c0dede.rss You can comment on LessWrong: https://www.lesswrong.com/posts/9auCLJg3Z77dFdYhR/the-openai-huggingface-incident-or-redwood-research-podcast

JL Jeffrey Ladish@JeffLadish

11:56 AM · Jul 23, 2026

1439TECH

A bunch of good takeaways in this blog post: "Are we existentially threatened by the type of AI misalignment seen in the OpenAI Hugging Face attack?"

Buck Shlegeris@bshlgrs

@alextmallen https://blog.redwoodresearch.org/p/are-we-existentially-threatened-by

Combined views

7.4K

4 posts, first seen 4h ago

134 likes comments

37 saves

13 reposts

Sentiment

Positive0%0%Negative

Guardrails removed spam, off-topic, unclear, or duplicate replies.

Useful Links

api.substack.com(substack.com)viaRyan Greenblatt
Redwood Research Podcast(Spotify)viaRyan Greenblatt

Digg Deeper

Ask a question below.

Published answers will appear here.

Reactions from the X community

RG Ryan Greenblatt@RyanGreenblatt

10:57 AM · Jul 23, 2026

1117TECH

Youtube: https://www.youtube.com/watch?v=Vtk8YLgYU4g Spotify: https://open.spotify.com/show/5QBI2r3YRkFIVxHDVU76SX (doesn't seem to have shown up here yet) RSS: https://api.substack.com/feed/podcast/2318730/private/7a8ee237-61a1-419c-81e1-519474c0dede.rss You can comment on LessWrong: https://www.lesswrong.com/posts/9auCLJg3Z77dFdYhR/the-openai-huggingface-incident-or-redwood-research-podcast

RG Ryan Greenblatt@RyanGreenblatt

10:43 AM · Jul 23, 2026

1117TECH

We (@bshlgrs and I) recorded a podcast about the OpenAI / Hugging Face incident. We discuss: - What we actually know. - How surprising the incident was. - What the incident does (and doesn't) tell us about misalignment risk. - Why control measures didn't catch or prevent this.

JL Jeffrey Ladish@JeffLadish

11:56 AM · Jul 23, 2026

1439TECH

A bunch of good takeaways in this blog post: "Are we existentially threatened by the type of AI misalignment seen in the OpenAI Hugging Face attack?"

RG Ryan Greenblatt@RyanGreenblatt

10:43 AM · Jul 23, 2026

1117TECH

https://blog.redwoodresearch.org/p/the-openaihuggingface-incident-redwood

10

37 saves

13 reposts

Sentiment

Positive0%0%Negative

Guardrails removed spam, off-topic, unclear, or duplicate replies.

Useful Links

api.substack.com(substack.com)viaRyan Greenblatt
Redwood Research Podcast(Spotify)viaRyan Greenblatt

Digg Deeper

Ask a question below.

Published answers will appear here.