Alessio Fanelli

speaker

10,045 appearances 92 recordings 1 series first heard Oct 2024 last heard 20 May

Alessio Fanelli’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 28 in all, peaking in Oct 2025 with 7.

Appearances

newest first · ▶ plays the moment

Latent Space: The AI Engineer Podcast · ⚡️Jailbreaking AGI: Pliny the Liberator & John V on Red Teaming, BT6, and the Future of AI Security · 16 Dec 2025 podcast

This is Alessio, founder of Kernel Labs, and I'm joined by Swix, editor of Laden Space. matched

Hello, hello. matched

How do you think about jail bricks overall? matched

So I think people understand the concept, but there's you know, some people then might say, Hey, are you jailbreaking to get instructions on how to make a bomb? matched

And I think that's what some of the, you know, people in in politics are trying to use to regulate some of the tech versus task specific jail bricks and things like that. matched

Just I think most people are not very familiar with like the scope of it. matched

So maybe just give people like a overview of like what it means. matched

means to like liberate a model. matched

Um and then we can kinda take it from there. matched

And can you maybe specify between jail breaking out of like a system prompt and, you know, more kind of like inference time security, so to speak, versus things that are being post trained out of the model and maybe the different levels of difficulty, like what is possible, what is not possible, and maybe the trajectory of the models, how better they've gotten. matched

I think the refusal is like one of the main benchmarks that the model providers still post and matched

GPD five point one, I think had like ninety-two percent refusal or something like that. matched

And then I think you joke broke in like one day. matched

I'm sure it it didn't take them one day to put the guy rails up, so it's pretty impressive the way you do it. matched

So maybe you walk us through that that process. matched

How how do you pick these dividers? matched

Like is there a science to it where like you're you know, taking the right word or like how much of it is like these are just my favorite tokens and they work for me and I bring them with me everywhere? matched

What do you think we should do to get more people to contribute open source data? matched

Like is it more bounties? matched

Is it yeah, I don't know. matched

Showing 1281–1300 of 10,045 · page 65 of 503