RSS

Elle Najt

Karma: 121

Prompted CoT Early Exit Un­der­mines the Mon­i­tor­ing Benefits of CoT Uncontrollability

Elle Najt, Asa Cooper Stickland and Xander Davies
17 Apr 2026 19:30 UTC
72 points
6 comments15 min readLW link

How Un­mon­i­tored Ex­ter­nal Agents can Sab­o­tage AI labs

Elle Najt and Fabien Roger
9 Apr 2026 18:07 UTC
23 points
0 comments9 min readLW link

Opus’s Schel­ling Steganog­ra­phy Has Am­plifi­able Se­crecy Against Weaker Eavesdroppers

Elle Najt7 Apr 2026 6:01 UTC
33 points
2 comments36 min readLW link

The Gold­borg Vari­a­tions: Al­go­rave At­trac­tor States of LLMs

Elle Najt1 Mar 2026 4:36 UTC
7 points
0 comments7 min readLW link