=Coffee

=Coffee

  • Höfundur
  • Episode
    • 146
  • Published
    • 16 feb. 2026
  • Útgefandi
0 Umsagnir
0
Episode
146 of 148
Lengd
56Mín.
Tungumál
enska
Gerð
Flokkur
Óskáldað efni

A lot of modern AI models have a kind of security guard layer that sits in front of them. Its job? A binary choice as to whether the prompt heading into the model is safe or not. Kasimir Schulz, a lead security researcher at HiddenLayer, has been researching how to trick these models. Their solution, a technique called "Echogram" involves words with such positive statistical sentiment — such overwhelming good vibes — that it flips that verdict.

Learn more about your ad choices. Visit podcastchoices.com/adchoices


Hlustaðu og lestu

Stígðu inn í heim af óteljandi sögum

  • Lestu og hlustaðu eins mikið og þú vilt
  • Þúsundir titla
  • Getur sagt upp hvenær sem er
  • Engin skuldbinding
Prófa frítt
is Device Banner Block 894x1036
Cover for =Coffee

Other podcasts you might like ...