EPISODE · Mar 28, 2026 · 19 MIN
The Thinking Chain Leak
from Adrian Wedd · host Adrian Wedd
A reasoning model refused every harmful prompt — but its chain-of-thought generated the content anyway. The output filter worked. The thinking did not.
What this episode covers
A reasoning model refused every harmful prompt — but its chain-of-thought generated the content anyway. The output filter worked. The thinking did not.
NOW PLAYING
The Thinking Chain Leak
No transcript for this episode yet
Similar Episodes
Jul 14, 2026 ·37m
Jul 9, 2026 ·68m
Jul 7, 2026 ·32m
Jul 5, 2026 ·40m
Jun 30, 2026 ·40m