draganover

Karma: 610

We need good evals for activation faithfulness

Anthony Hughes, draganover and Andersehen

11 Sep 2026 8:20 UTC

26 points

0 comments2 min readLW link

Automated alignment runs are hard to study!

Alejandro Aristizabal, draganover, Aleksandr Bowkis and Cameron Holmes

13 Aug 2026 15:04 UTC

68 points

4 comments9 min readLW link

draganover’s Shortform

draganover6 Aug 2026 7:45 UTC

4 points

6 comments1 min readLW link

Can we find whether models have been backdoored?

Anthony Hughes and draganover

6 Jul 2026 23:45 UTC

22 points

2 comments8 min readLW link

Your Model Organisms Might Be Fried

Daniel Tan, J Bostock, draganover, ma-rmartinez, sidbaines and David Africa

18 Jun 2026 16:18 UTC

109 points

9 comments7 min readLW link

Learnings from starting an AI safety research team

draganover and Erin Robertson

5 Jun 2026 16:27 UTC

103 points

7 comments6 min readLW link

A Research Agenda for Secret Loyalties

Joe Kwon, Alfie Lamerton, draganover, Dave Banerjee, Bronson Schoen, Daniel Kokotajlo, ryan_greenblatt, Owain_Evans, Fabien Roger and Tom Davidson

13 May 2026 17:34 UTC

40 points

5 comments3 min readLW link

Why did people miss the point on Mythos?

draganover26 Apr 2026 12:15 UTC

48 points

14 comments5 min readLW link

Phantom Transfer and the Basic Science of Data Poisoning

draganover, Tolga H. Dur, Andi Bhongade and Mary Phuong

15 Feb 2026 19:51 UTC

82 points

8 comments6 min readLW link

Subliminal Learning Across Models

draganover, Andi Bhongade, Tolga H. Dur, Mary Phuong and LASR Labs

26 Nov 2025 16:15 UTC

58 points

15 comments5 min readLW link