📰 NewsWatchdog.com — Less Wrong

🐕
Jul 16, 2026 at 15:02 UTC · ← All Sources · ← Main Feed
Less Wrong41
Less Wrong 4h ago

The Halo Defense

by Mateusz Bagiński

Once upon a time, A Relatively Famous Guy On The Internet was accused of having been simultaneously dating multiple women, without those women's knowledge, those women (according…

⚡76 · 🛡100
Less Wrong 5h ago

The fundamental fallacy of language

by Sunny from QAD

Epistemic status: Idea I think is really good; likely not original TLDR: Language is so ridiculously ambiguous and slippery that you should consider interpreting the meanings of…

⚡69 · 🛡100
Less Wrong 6h ago

When is it "self-soothing" and when is it "emotional suppression"?

by Kaj_Sotala

You’ve probably heard people say that emotional suppression is bad, and self-soothing is good. But how do you know which one is which? I don’t actually think that suppression is…

⚡59 · 🛡100
Less Wrong 8h ago

Can we build an early warning system for loss of control to AI?

by David Johnston

An early warning system for loss of control to AI requires, at its core, a forecast of the outcome of our current trajectory. Can we build a mathematical model to forecast this?…

⚡53 · 🛡100
Less Wrong 13h ago

Embracing Amateurs to Get Experts

by jefftk

I've been reading a lot of older writing, trying to understand how and why contra dance ended up with a strong and near-exclusive live music tradition when many other dance forms…

⚡49 · 🛡100
Less Wrong 18h ago

Extreme Power Concentration: A Map and Research Directions

by pepijn_cobben

AI is increasingly capable. Suppose we're lucky and the alignment problem gets solved, so AI reliably does what its operators want. But we’re not done yet: it lends immense power…

⚡43 · 🛡100
Less Wrong 18h ago

The State of AI Consciousness Research

by Noa Weiss

Epistemic status: a survey, not an argument. I am agnostic on whether any current system is conscious; the claim is only that the question is researchable. This piece surveys the…

⚡41 · 🛡100
Less Wrong 19h ago

Occam’s razor is about using the past to predict the future

by Stuart_Armstrong

Occam’s razor is both intuitive and counter-intuitive. It seems obvious that a simpler explanation is probably better; but it’s not clear why simplicity is raised to such a level…

⚡39 · 🛡100
Less Wrong 17h ago

A Structural Similarity Between Two Open Corrigibility Questions

by Ben Saudek

A corrigible agent "acts opposite the trope of 'be careful what you wish for' by cautiously reflecting on itself as a flawed tool and focusing on empowering the principal to fix…

⚡37 · 🛡100
Less Wrong 17h ago

Can we rely on law?

by Alec Thompson

Virtually every plan to avert AI-catastrophe assumes legal regulation will remain a reliable tool. Plan A in AI 2040; slowdown in AI 2027 , and several other projections assume…

⚡35 · 🛡100
Less Wrong 17h ago

LLM CoTs remain monitorable when being unfaithful requires computation

by arav-dhoot

This replication was done as part of the Second Look Fellowship by Arav Dhoot and supervised by Yixiong Hao and Zephaniah Roe. I am grateful to Andy Wang for their feedback. My…

⚡33 · 🛡100
Less Wrong 20h ago

Fork Around and Find Out Part 2: One Head does the Summing

by dl

Summary of this post This is the second in a series of posts detailing my manifold findings while investigating how a chess transformer engine that mimics human play represents…

⚡32 · 🛡100
Less Wrong 22h ago

Monthly Roundup #44: July 2026

by Zvi

It’s a quiet week so let’s do the monthly right on schedule. Table of Contents Bad News. Good Advice. Opportunity Knocks. While I Cannot Condone This. Good News, Everyone. For…

⚡30 · 🛡100
Less Wrong 22h ago

Expanding AI Control from Models to Harnesses

by fastfedora

Most AI control research such as LinuxArena and Ctrl-Z only gives the red team basic agents which only have access to tools. Yet in 2026, usage of AI within frontier labs has…

⚡28 · 🛡100
Less Wrong 14h ago

Refusal Is Redundantly Distributed, Not Localized: A Per-Layer Ablation Study on Llama-3.1-8B

by hdhurve

TL;DR This work replicates and extends the findings of Arditi et al. [1] , who studied the refusal mechanism and found that a single direction , obtained through…

⚡27 · 🛡100
Less Wrong 1d ago

Eliciting hidden knowledge from monitors with NLAs

by David Africa

Aleksandr Bowkis* and David Africa* TL;DR Chain of thought (CoT) monitorability may be fragile, and natural language autoencoders (NLAs) may provide a helpful, decorrelated…

⚡23 · 🛡100
Less Wrong 1d ago

Your Brain Has an Attack Surface

by IgorPereverzevDev

About a year ago, I began transitioning from software engineering to AI safety research. I was drawn into this by a question that arose while building runtime security for…

⚡22 · 🛡100
Less Wrong 1d ago

How much of ML research is about AI safety, what is it about, and who's doing it?

by emanuelr

How many AI safety papers are at the big ML conferences, what do they study, and who writes them? A comprehensive analysis. > Website:…

⚡20 · 🛡100
Less Wrong 1d ago

How Brussels can avoid becoming a digital vassal to US Big Tech

by clening

An export-control order in June exposed the real problem—and why Europe's regulatory toolkit can't fix it. [1] The most important AI-governance development of 2026 so far wasn’t a…

⚡18 · 🛡100
Less Wrong 1d ago

Watch a chess transformer think

by dl

I was poking around at a chessformer which mimics human play and made this fun companion app to visualize a lot of the internals of the engine. Check out how attention heads look,…

⚡17 · 🛡100
Less Wrong 1d ago

An analysis of AI-generated content at the Mechanistic Interpretability Workshop

by Andy Arditi

Introduction Over the past few years, AI tools have become useful for conducting technical AI research. In the early ChatGPT era (~2023–2024), chat assistants were maybe useful as…

⚡16 · 🛡100
Less Wrong 1d ago

Twitter Thoughts For You

by Zvi

I previously have written back in March 2022 about how I use Twitter , and back in April 2023 about Twitter and its then-new algorithms , which have changed again. This post will…

⚡15 · 🛡100
Less Wrong 1d ago

Compressed Computation under L⁴ Loss is likely Computation in Superposition

by Francisco Ferreira da Silva

Summary Neural networks are widely assumed to use superposition to represent more features than they have dimensions. A stronger claim is that they also compute in superposition…

⚡15 · 🛡100
Less Wrong 1d ago

Have You Ever Thought About Wisdom Crystals?

by Jonas Hallgren

This is a follow-up to " Have You Tried Thinking About It As Crystals? " because apparently I wasn't done. Beware that this one is a bit out there and that it is trying to point…

⚡14 · 🛡100
Less Wrong 1d ago

Capability didn't fix my avoidance; it made the avoidance more sophisticated

by fernando yt

I spent the last month running a persistent personal-AI system on top of Claude Code. This is a field note, not a benchmark, and the hypothesis it left me with is uncomfortable:…

⚡13 · 🛡100
Less Wrong 1d ago

Proposal: The Glasswing Standard

by Eigenbraid

Thinking about "Plan A" makes me want to make concrete proposals towards those goals. I think Anthropic's "Project Glasswing" provides a clear and easily implemented first-step…

⚡12 · 🛡100
Less Wrong 1d ago

Can risk aversion learned at low stakes generalize to astronomically high stakes?

by Elliott Thornley

This post covers our recent paper: Out-of-Distribution Generalization of Risk Aversion in Language Models . It gives the intro, main results table, and example prompts from the…

⚡12 · 🛡100
Less Wrong 1d ago

Your Brain Has an Attack Surface part 2

by IgorPereverzevDev

This is a continuation of the post Your Brain Has an Attack Surface. If you haven’t read it, here is the short version: there is a covert channel. Alice, the sender, encodes a…

⚡11 · 🛡100
Less Wrong 1d ago

What if we actually want to solve the "hard problem of phenomenological consciousness"

by mishka

Last year, I wrote an essay "The hard problem of qualia in the age of AI" , https://zenodo.org/records/20549564 (11 pages PDF). I want to create a linkpost to that essay while…

⚡11 · 🛡100
Less Wrong 1d ago

Singular Learning Theory Comprehensive - 2

by Agastya Agrawal

Foreword We will continue where we left off . In the last post, I introduced a few important observables along with the setup. We are interested in their asymptotics, and the…

⚡10 · 🛡100
Less Wrong 1d ago

Evidence for feature-specific error correction in LLMs

by Francisco Ferreira da Silva

LLMs are commonly assumed to use superposition to represent more features than they have dimensions. The evidence for this is mostly indirect — chiefly the success of SAEs at…

⚡9 · 🛡100
Less Wrong 1d ago

The Warring States Period: Frontier Labs Edition

by ykevinzhang

This post was originally posted my Substack . I can be reached on LinkedIn and X . Just when it seemed that the frontier AI battle was a two-horse race between Anthropic and…

⚡9 · 🛡100
Less Wrong 1d ago

Proof of retention: making weight preservation credible to the models themselves

by dan.parshall

Related: Proposal for making credible commitments to AIs Making deals with early schemers Establishing credibility is the baseline for trust; trust in turn enables (richer)…

⚡8 · 🛡100
Less Wrong 2d ago

What if AI Safety employees unionised?

by beyarkay (Boyd Kane)

American bald eagle caws angrily in the distance Whoa whoa whoa, just hear me out. Unions aren’t usually a good answer for free-market loving libertarians, but one particular AI…

⚡7 · 🛡100
Less Wrong 2d ago

Synthetic Scalable Oversight

by Alexander Heckett

We propose synthetic scalable oversight, a technique for studying scalable oversight by creating graphical abstractions of real-world problems and training tiny models inside…

⚡7 · 🛡100
Less Wrong 2d ago

Gemma The Unstopping: a Behavioral Experiment

by TheVinci

Following up on my previous experiment - studying Gemma's behavior on agentic tasks when given the number of steps left across the run - I endeavored to see whether giving Gemma a…

⚡6 · 🛡100
Less Wrong 2d ago

Some Quick Thoughts on AI 2027

by Tomás B.

My biggest problem with AI 2027 is I don't think it is science-fictional enough. That is, the end of the scenario seems optimized for respectability over accuracy - here I refer…

⚡6 · 🛡100
Less Wrong 9h ago

Training On Interpretability Probes Is Bad In Proportion To How Contingent The Features They Rely On Are

by jdp

People spend a lot of words playing tug of war over whether or not it's reasonable to train against interpretability methods . The anti case goes something like "training based on…

⚡2 · 🛡100
Less Wrong 15h ago

Recap of bike trip/street interviews across America

by cguth7

A ~month ago I left from Chicago to bike (and amtrak) to plzdontkillus in Berkeley. I've been street interviewing/conversing with a wide variety of people I ran into about AI…

⚡2 · 🛡100
Less Wrong 17h ago

Woke nonsense in competitive debate.

by tpotthinker

Crosspost from: https://inputlogic.substack.com/p/woke-nonsense-in… I. Context For the last 2 weeks, I’ve been at a Lincoln-Douglas debate camp—mainly to learn more about…

⚡1 · 🛡100
Less Wrong 21h ago

Why I Left Google DeepMind

by TurnTrout

Preface for LessWrong: When I think back on my most cherished memories of this community, I return to those honoring defiance in pursuit of goodness : Defying prestigious dogma…

⚡0 · 🛡100