RSS

Pro­ject Glass­wing: An­thropic Shows The AI Train Isn’t Stopping

AlphaAndOmega7 Apr 2026 21:02 UTC
11 points
0 comments4 min readLW link

Role-play­ing vs Self-modelling

Jan_Kulveit7 Apr 2026 20:41 UTC
18 points
0 comments4 min readLW link

Claude Mythos Sys­tem Card Preview

anaguma7 Apr 2026 20:29 UTC
44 points
1 comment3 min readLW link
(www-cdn.anthropic.com)

The Train­ing Ex­am­ple Lie Bracket

DaemonicSigil7 Apr 2026 20:13 UTC
6 points
0 comments1 min readLW link
(pbement.com)

A con­ver­sa­tion with An­ima Labs, part I: Phenomenol­ogy of digi­tal minds

7 Apr 2026 19:19 UTC
11 points
0 comments37 min readLW link
(smoothbrains.net)

Fan­tasy ideology

Ninety-Three7 Apr 2026 17:52 UTC
22 points
0 comments11 min readLW link

My pic­ture of the pre­sent in AI

ryan_greenblatt7 Apr 2026 16:44 UTC
65 points
5 comments11 min readLW link

Beliefs are Cho­sen to Serve Goals

Ashe Vazquez Nuñez7 Apr 2026 16:43 UTC
23 points
1 comment4 min readLW link
(tuesdaybornwhale.substack.com)

An Align­ment Jour­nal: Fea­tures and policies

7 Apr 2026 15:22 UTC
19 points
0 comments15 min readLW link
(blog.alignmentjournal.org)

We’re ac­tu­ally run­ning out of bench­marks to up­per bound AI capabilities

LawrenceC7 Apr 2026 6:47 UTC
49 points
4 comments4 min readLW link

“Align­ment” and “Safety”, part one: What is “AI Safety”?

David Scott Krueger (formerly: capybaralet)7 Apr 2026 6:10 UTC
15 points
1 comment2 min readLW link
(therealartificialintelligence.substack.com)

Opus’s Schel­ling Steganog­ra­phy Has Am­plifi­able Se­crecy Against Weaker Eavesdroppers

Elle Najt7 Apr 2026 6:01 UTC
26 points
0 comments36 min readLW link

My Ethics

NickyP7 Apr 2026 4:18 UTC
8 points
7 comments6 min readLW link
(blog.sus.cat)

Don’t write for LLMs, just record everything

RobertM7 Apr 2026 3:12 UTC
37 points
4 comments6 min readLW link

Vibe an­a­lyz­ing my genome

Ruby7 Apr 2026 3:05 UTC
5 points
3 comments11 min readLW link

To­ken-Level Fork­ing Paths in Rea­son­ing Traces: Some Examples

Rob D7 Apr 2026 2:37 UTC
3 points
0 comments58 min readLW link

By Strong De­fault, ASI Will End Liberal Democracy

MichaelDickens6 Apr 2026 23:43 UTC
37 points
12 comments3 min readLW link

The Garden

sturb6 Apr 2026 22:35 UTC
4 points
0 comments5 min readLW link
(www.benjaminsturgeon.com)

Con­tra Nina Pan­ickssery on ad­vice for children

Sean Herrington6 Apr 2026 21:41 UTC
52 points
3 comments3 min readLW link

Are there Mul­ti­ple Mo­ral End­points?

Vaniver6 Apr 2026 20:37 UTC
21 points
5 comments5 min readLW link