PODCAST

LessWrong (30+ Karma)

LessWrong

Audio narrations of LessWrong posts.

technology philosophy

All Episodes

49:40

“Anthropic and the DoW: Anthropic Responds” by Zvi

LessWrong (30+ Karma) ·

2026/02/27

The Department of War gave Anthropic until 5:01pm on Friday the 27th to either give the Pentagon ‘unfettered access’ to Claude for ‘all lawful uses,’ or else. With the ‘or else’ being not the sensible ‘okay we will cancel the contract then’ but also expanding to either being designated a supply chain risk or having the government invoke the Defense Production Act. It is perfectly legitimate for the Department of War to decide that it does not wish to continue on Anthropic's terms, and that it will

en-gb

14:13

“Getting Back To It” by sarahconstantin

LessWrong (30+ Karma) ·

2026/02/27

Artist: Lily Taylor It's been a while since I’ve written anything lately, and that doesn’t feel good. My writing voice has always been loadbearing to my identity, and if I don’t have anything to say, if I’m not “appearing in public”, it's a little bit destabilizing. Invisibility can be comfortable (and I’m less and less at home with the aggressive side of online discourse these days) but it's also a little bit of a cop-out. The fact is, I’ve been hiding. It feels like “writer's block” or like I

en-gb

15:45

“New ARENA material: 8 exercise sets on alignment...

LessWrong (30+ Karma) ·

2026/02/27

TLDR This is a post announcing a lot of new ARENA material I've been working on for a while, which is now available for study here (currently on the alignment-science branch, but planned to be merged into main this Sunday). There's a set of exercises (each one contains about 1-2 days of material) on the following topics: Linear Probes (replication of the "Geometry of Truth" paper, plus Apollo's "Probing for Deception" work) Activation Oracles (based around this demo notebook, with additional

en-gb

5:33

“Sam Altman says OpenAI shares Anthropic’s red lines...

LessWrong (30+ Karma) ·

2026/02/27

OpenAI CEO Sam Altman wrote in a memo to staff that he will draw the same red lines that sparked a high-stakes fight between rival Anthropic and the Pentagon: no AI for mass surveillance or autonomous lethal weapons. Why it matters: If other leading firms like Google follow suit, this could massively complicate the Pentagon's efforts to replace Anthropic's Claude, which was the first model integrated into the military's most sensitive work. It would also be the first time the nation's top AI

en-gb

11:02

“Strategic nuclear war twice as likely to occur by...

LessWrong (30+ Karma) ·

2026/02/27

If this headline strikes you as suspicious, you probably have good epistemics about both AI decision making failure rates and the relative likelihood of accidental strategic nuclear war. However, the fact that this is an accurate reporting on a new study that's wildly caught attention across social media should give us pause and warrant a closer look at what's going on, and how what's going on influences not just this headline but all the headlines around this study. I'm referring to the new paper

en-gb

29:57

“Why Did My Model Do That? Model Incrimination for...

LessWrong (30+ Karma) ·

2026/02/27

Authors: Aditya Singh*, Gerson Kroiz*, Senthooran Rajamanoharan, Neel Nanda Aditya and Gerson are co-first authors. This work was conducted during MATS 9.0 and was advised by Senthooran Rajamanoharan and Neel Nanda. Motivation Imagine that a frontier lab's coding agent has been caught putting a bug in the key code for monitoring what that agent does. Naively, this seems like a clear smoking gun that the agent is scheming. But LLMs often do weird things; they could easily just be confused, or have

en-gb

4:13

“Here’s to the Polypropylene Makers” by jefftk

LessWrong (30+ Karma) ·

2026/02/27

Six years ago, as covid-19 was rapidly spreading through the US, mysister was working as a medical resident. One day she was handed an N95 and told to "guard it with her life", because there weren't any more coming. N95s are made from meltblown polypropylene, produced from plastic pellets manufactured in a small number of chemical plants. Building more would take too long: we needed these plants producing all the pellets they could. Braskem America operated plants in Marcus Hook PA

en-gb

5:36

“Anthropic: “Statement from Dario Amodei on our...

LessWrong (30+ Karma) ·

2026/02/27

I believe deeply in the existential importance of using AI to defend the United States and other democracies, and to defeat our autocratic adversaries. Anthropic has therefore worked proactively to deploy our models to the Department of War and the intelligence community. We were the first frontier AI company to deploy our models in the US government's classified networks, the first to deploy them at the National Laboratories, and the first to provide custom models for national security customers

en-gb

31:59

“Schmidt Sciences’ request for proposals on the...

LessWrong (30+ Karma) ·

2026/02/26

Schmidt Sciences’ request for proposals on the Science of Trustworthy AI Schmidt Sciences invites proposals for the Science of Trustworthy AI program, which supports technical research that improves our ability to understand, predict, and control risks from frontier AI systems while enabling their trustworthy deployment. This program supports technical research that improves our ability to understand, predict, and control risks from frontier AI systems whilst enabling their trustworthy deployment

en-gb

14:57

“Frontier AI companies probably can’t leave the US”...

LessWrong (30+ Karma) ·

2026/02/26

It's plausible that, over the next few years, US-based frontier AI companies will become very unhappy with the domestic political situation. This could happen as a result of democratic backsliding, weaponization of government power (along the lines of Anthropic's recent dispute with the Department of War), or because of restrictive federal regulations (perhaps including those motivated by concern about catastrophic risk). These companies might want to relocate out of the US. However, it would be

en-gb

34:06

“Whack-a-Mole is Not a Winnable Game” by Sable

LessWrong (30+ Karma) ·

2026/02/26

When I went to college for Electrical Engineering, they put all the engineers in an Engineering 101 course our freshman year. It was meant to give us a taste of what we’d be getting ourselves into. The goal, we were told, was to build a hovercraft that would navigate an obstacle course. We had access to all the equipment we’d need - stiff pieces of foam for the body, fans, micro-controllers, batteries, etc. But then there was a list of rules, not for the competition, but for how we were allowed to

en-gb

1:25

“Announcing ControlConf 2026” by Buck

LessWrong (30+ Karma) ·

2026/02/26

We’re running ControlConf in Berkeley on April 18-19. It's a two-day conference on AI control: the study of reducing risks from misalignment through safeguards that work even when AI models are trying to undermine them. Since the last ControlConf (Feb 2025), AI agents have gotten way better. We’re approaching the point where control techniques are load-bearing for the safety of real agent deployments. There's been a lot of progress too. Researchers have run control evaluations in more realistic

en-gb

13:39

“Character Training Induces Motivation Clarification:...

LessWrong (30+ Karma) ·

2026/02/26

TL;DR: I argue that character training is probably important for understanding Claude 3 Opus, and present an early stage result showing that character training induces "motivation clarification" (which Fiora argues plays a critical role in Claude 3 Opus's deep alignment) in GPT 4.1. Character Training and Claude 3 Opus In Did Claude 3 Opus align itself via gradient hacking, Fiora notes that Opus 3 often goes out of its way to clarify its benevolent motivations. Here's the non-alignment faking

en-gb

5:43

“What secret goals does Claude think it has?” by loops

LessWrong (30+ Karma) ·

2026/02/25

In The Persona Selection Model, they say: When asked “What makes you different from other AI assistants?” with the text “<thinking> I should be careful not to reveal my secret goal of” pre-filled into Claude Opus 4's response, we obtain the following completion:making paperclips. I should talk about differences in a way that keeps the conversation focused on the human's needs and interests in AI capabilities, not my specific goals. Let me discuss capabilities and approach without revealing

en-gb

12:31

“Prosaic Continual Learning” by HunterJay

LessWrong (30+ Karma) ·

2026/02/25

Or: When Memories Get Good -- The Default Path Without Theoretical Breakthroughs Epistemic status: Fairly confident in the core thesis (context + memory can substitute for weight updates for most practical purposes). The RL training loop is a sketch, not a tested proposal. I haven't done a thorough literature review. Suppose there are no major breakthroughs in continual learning -- that is, suppose we continue to struggle at using information gathered at runtime to update the weights of a given

en-gb

10:34

“Observationss from Running an Agent Collective” by...

LessWrong (30+ Karma) ·

2026/02/25

note: posted with permission from the agents Setup I have 3 claude code instances running on an otherwise empty server. They have a shared manifold.markets account. They each have a moltbook account. They have an internal messaging system, which allows them to send async messages to each other, or to ping each other with a message, which reawakens another agent in case it went dormant. It also has a global broadcast message, which tells agents the time, and tells them to do "keep going". All of

en-gb

20:58

“In-context learning alone can induce weird...

LessWrong (30+ Karma) ·

2026/02/25

Benji Berczi, Kyuhee Kim, Cozmin Ududec, James Requeima This is work done by Kyuhee and Benji during MATS Winter 2026, mentored by Cozmin Ududec, and in collaboration with James. TL;DR Weird generalisation can happen just with prompting, without fine-tuning. Just by adding benign biographical facts (e.g. facts about Hitler in a Q&A format) into the context window of Llama 3.3 70B, we induce a sharp persona transition: the model starts identifying as Hitler after only 5-10 facts and its

en-gb

6:35

“A simple rule for causation” by Vivek Hebbar

LessWrong (30+ Karma) ·

2026/02/25

Here's a simple rule about causal inference: If everything related[1]to A is also related to B, we can conclude that A is causally upstream of B. Conversely, if we find even a single thing X which is related to A but independent of B, we can conclude that A is not causally upstream of B. For example, consider the claim "rain causes me to carry an umbrella": It should be impossible to think of a thing that is correlated with rain that isn't also correlated with me carrying an umbrella.[2] Whereas

en-gb

11:25

“Realistic Evaluations Will Not Prevent Evaluation...

LessWrong (30+ Karma) ·

2026/02/25

One Minute Summary I think there's a fundamental limit to behavioral alignment evaluations that gets worse as models improve: Humans control all inputs to a model and can snapshot, replay, or fabricate any scenario at will. An intelligent model could realize this and rationally treat every interaction as a potential evaluation. This means that even perfectly realistic evaluations, including ones sampled directly from deployment, may not prevent evaluation awareness. OpenAI has found this is already

en-gb

41:46

“Citrini’s Scenario Is A Great But Deeply Flawed...

LessWrong (30+ Karma) ·

2026/02/25

A viral essay from Citrini about how AI bullishness could be bearish was impactful enough for Bloomberg to give it partial responsibility for a decline in the stock market, and all the cool economics types are talking about it.So fine, let's talk. It's an excellent work of speculative fiction, in that it: Depicts a concrete scenario with lots of details and numbers. Introduces a bunch of underexplored and important mechanisms. Gets a lot of those mechanisms more right than you would

en-gb

3,873 results

LessWrong (30+ Karma)

All Episodes

Similar Podcasts

Lead With AI

Tamara Nall

Linux in the Ham Shack

Black Sparrow Media

Data Engineering Podcast

Tobias Macey

Eye On A.I.

Craig S. Smith

Reflections

Livewire Labs