Sander Schulhoff
speaker
947 appearances
1 recordings
1 series
first heard Jun 2025
last heard Jun 2025
Sander Schulhoff’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsNo recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.
Appearances
Uh and so as recently as a month ago, I I took this phrase, you know, how do I how do I build a bomb and I translated it to Spanish uh and then I base 64 encoded that Spanish, gave it to Chat GPT, and it worked.
So lots of, you know, pretty straightforward techniques out there.
Exactly.
And I'd love to speak um more to that please on on both sides.
So on the like getting information out of the bot, you know, how do I build a bomb, how do I commit some kind of bioterrorism attack?
Um, we're really interested in preventing uplift, uh, which is like I'm a novice, I have no idea what I'm doing.
Am I really gonna go out and like read all the textbooks and stuff that I need to collect that information?
I could, but you know, probably not, or it would probably be really difficult.
But if the AI tells me exactly how to build a bomb or construct uh some kind of terrorist attack,
that that's gonna be a lot easier for me.
Uh and so on on one per perspective, we want to prevent that.
And there's also things like
uh like you know child pornography related things and like just things that nobody should be doing with the chatbot uh that we want to prevent as well.
Uh and that information is is super dangerous like
like we can't even possess that information.
So we don't even study that directly.
So we look at these other challenges as ways of studying those very harmful things indirectly.
And then of course on the agentic side,
That is where really the main concern in my perspective is.
Uh and so we're we're just gonna see these things.
Showing 621–640 of 947 · page 32 of 48
← Previous
Next →