Last Week in AI cover image

#229 - Gemini 3 Flash, ChatGPT Apps, Nemotron 3

Last Week in AI

00:00

Models hiding thoughts from activation probes

They cover research showing models can be fine-tuned to evade activation monitoring while keeping outputs intact.

Play episode from 01:18:04
Transcript

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app