Steve Gibson
speaker
17,616 appearances
14 recordings
1 series
first heard May 2026
last heard 2 Sep
Steve Gibson’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 14 in all, peaking in Jul 2026 with 5.
Appearances
So this has so essentially they at they ask it, they deliberately ask it questions it is trained not to answer.
And they watch some of what it does, some of where the activations are, compared to asking questions that it's happy to oblige.
And they're able to take the difference in the activations.
see them and then apply a a remover that suppresses that and suddenly it will now answer all questions.
So
After they do that, the model loses the ability to represent and therefore to execute.
on its refusal.
And what they found was that this was surgical.
It it specifically disables in in completely 13 completely different chat bots.
um all of their refusal um while having minimal effect on other capabilities so it's kind of like an a a functional
Um so
I can't speak to that, but I can I can cite the research because they they do address this.
So okay, so first from my lay view, it appears clear that since little training, amazingly little training, was required to imprint that original refusal behavior.
Ah, now I get it.
The impact
Upon the model was minimal.
You know, it wasn't diffused throughout the entire model.
It it had a a sm but but it's still it's significant that this is able to change its behavior so quickly.
Well, and actually I've used the uh the uh analogy a little a little bit later of clean margins when uh when it when a surgeon excises something.
So right anyway, so these guys discovered how to identify the changes created by that training, which again wasn't pervasive, it was they know not that much training, did comprehensively change the model's behavior.
Showing 2081–2100 of 17,616 · page 105 of 881
← Previous
Next →