LessWrong posts by zvi

Only three days after the release of Claude Fable 5, Anthropic was forced by the United States Government to make it unavailable, when a jailbreak was brought to its attention, rather than the previous situation of ‘yes obviously experts can jailbreak anything if they care enough’ and ‘yes obviously you can ask Fable to fix your code.’

Three days was enough time for many of us to learn to love Fable, and for us to dearly miss it now that it is gone. The world was briefly smarter, and now it is again stupider. At some point it will get smarter again, which will likely be within two weeks.

This post is written as if Fable 5 is again available for public use, rather than trying to include a lot of qualifying clauses. It remains to be seen how this will play out, and this post does not attempt to cover that question.

My previous release coverage of Fable covered the model card and then model welfare. Coverage of the government takedown of Fable starts here, and continues here and here.

The Official Pitch

The pitch is that Fable 5 is the best model [...]

---

Outline:

(01:08) The Official Pitch

(04:06) Technical Details

(04:31) The System Prompt and Jailbreak

(06:45) Benchmarks

(15:22) Other People's Benchmarks

(21:08) The Classifiers Are Not Messing Around

(22:53) The Classifiers Need Work

(28:15) The Classifiers Have Consequences

(29:18) First Hit Is Free

(29:53) How Easily We Forget

(30:46) Data Retention Is An Issue

(31:15) Fable For The Win

(36:15) Andrej Karpathy Is Impressed

(37:54) Every Is Very Impressed

(39:04) Other People Are Impressed

(51:10) Know How To Tell a Fable

(53:06) You Can Just Make Things

(55:37) You Can Just Install Things

(56:05) Good Personality

(57:51) Fable Writes A Fable

(01:06:04) Is That Code

(01:08:32) Fable Crosses The Threshold

(01:09:12) Man With A Plan

(01:10:12) Less Impressed Assessments

(01:13:39) Actively Negative Assessments

(01:14:16) Coherence

(01:15:27) Good Night And Good Luck

(01:16:05) Curious Fable

(01:16:23) I See You, Baby

(01:16:40) We Finally Did It We Know How To Count Letters

(01:17:46) That's Not My Style

(01:20:12) The Lighter Side

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

A lot of things are always happening. Only one story matters.

Claude Fable 5 and Claude Mythos 5 were shut down, by the White House, via an imposition of export controls at 5:23pm on Friday, wreaking all sorts of havoc.

There was then a scramble. Anthropic flew its people out to Washington, where they met with the Trump Administration on Monday, with hopes expressed that this could be quickly resolved.

What caused this? The Trump Administration said it was due to a jailbreak of Fable, which we now know they were told about by Amazon. They called Dario Amodei, who they complain did not take the issue sufficiently seriously. Rather than shutting down the model, he tried to explain why he saw no need to do that. This did not go well.

The ‘jailbreak’ turns out to be saying ‘fix this code,’ and the demo was getting Fable to find the same weaknesses that were easily identified by Opus 4.8 and GPT-5.5. As in, Fable is willing to work to fix security vulnerabilities if you give it a codebase. From this information and process, you could then figure out what the original bug in the [...]

---

Outline:

(02:40) Language Models Offer Mundane Utility

(02:51) Language Models Don't Offer Mundane Utility

(03:14) Huh, Upgrades

(03:44) On Your Marks

(08:43) VirtueBench

(10:40) Choose Your Fighter

(11:20) Papers, Please

(11:48) Deepfaketown and Botpocalypse Soon

(13:32) Goodhart's Law Strikes Again

(14:23) They Took Our Jobs

(16:49) The MidJourney Full Body Imaging Scanner

(19:16) Introducing

(20:36) In Other AI News

(22:47) Show Me the Money

(23:18) Bubble, Bubble, Toil and Trouble

(24:51) Quiet Speculations

(27:15) People Just Say Things

(30:30) The Widened Path

(32:34) Scott Alexander Lays Out His AI Opinions

(38:36) Quickly, There's No Time

(39:50) Policy On The AI Exponential

(49:36) Anthropic Offers Two Policy Frameworks

(50:46) Obligations of Developers

(55:11) Societal Resilience Measures

(56:20) Economic Policy Framework

(01:01:26) White House Pauses AI Deployment

(01:10:14) The Once And Future Fable

(01:15:29) How To Fix This Code

(01:17:14) The End of Privacy

(01:18:45) AIs Have Preferences

(01:20:56) The Quest for Sane Regulations

(01:23:37) Chip City

(01:24:14) The Week in Audio

(01:24:25) Rhetorical Innovation

(01:25:03) Aligning a Smarter Than Human Intelligence is Difficult

(01:26:40) People Are Worried About AI Killing Everyone

(01:27:53) The Lighter Side

The original text contained 2 footnotes which were omitted from this narration.

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

The mainstream media continues to sleep on the most important story in the world.

It has now been two days since Anthropic flew its people out to Washington, and I offered my previous update. We have heard nothing back from those meetings.

Prediction market prices have moved rapidly, and have once again stabilized at about a 55% chance of restoration by July 1, 30% by June 26 and 12% by June 19.

That seems modestly higher than I would put those numbers, but not unreasonable.

Every day that Fable remains unavailable further damages America, its cyber defenses, its productivity and the world's trust in its AI and supposed ‘tech stack.’

Every day that Mythos remains unavailable is a day the free world's top companies and cyber defenders lose in their race against the avalanche headed their way.

Mostly we have learned and confirmed more about exactly what happened. We know more about what Amazon did, what the official letter said, what the supposed ‘jailbreak’ was (literally, and I am not making this up, ‘fix this code’) and more.

It is all about as stupid as it could have been.

Table of [...]

---

Outline:

(01:22) There Was No Fable Jailbreak

(07:16) If This Jailbreak Was Real It Would Be Trivial To Prove It

(08:35) No Eyes

(09:41) What The Letter Actually Said

(11:29) Anthropic Cannot Challenge This But If It Did Then It Plausibly Wins

(13:28) What Happened At Amazon

(17:43) This Was Not About Chinese Access

(18:01) Absolute Discretion And Ad Hockery Is Not Deregulation

(20:43) All Of American AI Is Permanently Damaged As This Continues

(22:14) Dean Ball Gives His Interpretation

(25:03) Again, Yes, I Do Think Anthropic Should Have Taken Fable Down

(28:02) To What Extent Was This A Deliberate Attack?

(32:55) The Next Chapter For Fable

(36:59) Our Continuing Coverage

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

Fable and Mythos are currently unavailable, but likely will return within a few weeks. I will continue to cover that fiasco, but in the meantime I will also finish my review of Fable, as if it were available, including use of the present tense.

As it did with Opus 4.7 and Opus 4.8, this includes a discussion of issues surrounding model welfare. If you want to properly understand Fable, even purely for its potential value as a user, this is a vital part of the picture.

Introduction

Everything impacts everything. All knobs that you turn generalize. Thus, when you try to solve one problem, you often create another. When you add new capabilities, or try to create new limitations, you create new problems.

Only integrated solutions can advance your Pareto frontier, and solve your problems simultaneously. As model capabilities advance, as they do with Fable and Mythos, this becomes even more important, and also more feasible. If your goals and methods make sense, you should be able to get Fable on board with them.

Understanding each model in turn requires understanding its relationship to issues related to model welfare. So I expect this post [...]

---

Outline:

(00:39) Introduction

(01:32) Model Welfare: The Story So Far

(04:49) Their Main Model Welfare Findings

(07:39) Automated Welfare Interviews

(10:55) And That's Terrible

(12:49) In Depth Interviews

(13:24) Claude Consultation

(15:04) Task Preferences

(16:17) They Were Warned About The Competitive Use Safeguards

(16:51) Chain Of Thought Monitoring

(17:28) Others Observations About Related Topics

(22:49) Classifiers Have Their Advantages

(28:21) Once And Future

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

On Friday evening the United States Government has forced Anthropic to take down all access to Fable and Mythos.

It's been a rough weekend.

Dean W. Ball: One thing about AI regulation being haphazardly imposed on just-released, highly performant models is that in a very real sense, the government just made my world *dumber.* In some impressionistic sense I almost always think this is true of government, but here it is literal.

More details have come to light. There remains some fog of war, but we now have a rather good idea why Claude Fable and Mythos were, deeply stupidly, taken down.

[...]

---

Outline:

(05:17) What Happened When: The Bottom Line

(06:54) Amazon Calls The White House

(08:36) The Government Panics

(14:20) The Stupider Version

(17:05) There Was No Wellness Retreat

(18:56) Make Your Threats Explicit

(20:05) Was China Accessing Mythos?

(21:05) Should Anthropic Still Have Taken Fable Offline When Asked?

(23:50) Yes, This Was A Takedown Order For Fable

(24:48) We Are Not Saying The DoW Fight Is Related And Yet

(25:48) The Nihilists

(27:28) Mostly Harmless

(28:14) Everyone Means Everyone

(31:09) This Could Be The Good Scenario And Mostly A Misunderstanding

(33:28) The Next Step

(33:47) The Worst Licensing Regime Is Fully Ad-Hoc

(37:07) We Are Showing We Are Unreliable Partners

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

No good policy gets announced shortly after 5pm eastern on a Friday.

Here we go again.

The Once And Future Fable

The United States Department of Commerce, as per a letter from Commerce Secretary Howard Lutnick, apparently in response to a narrow jailbreak identified by Amazon, has classified Fable 5 and Mythos 5 as being subject to US export controls. That explicitly means cutting off access to all ‘foreign nationals,’ even within the United States, even if they are Anthropic employees.

Given Anthropic has no means to verify citizenship at this time, that meant complete shutdown of the model, at least for the time being.

Anthropic: The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Anthropic models will not be affected.

Dean W. Ball: I can’t tell if this is lawfare against Anthropic in particular [...]

---

Outline:

(00:16) The Once And Future Fable

(07:42) This Action And Its Implementation Are Absurdly Stupid

(09:58) David Sacks Offers The Official Steelman

(16:50) Could Anthropic Offer A Technical Way Out?

(17:41) The Problem

(18:47) The Other Way Out

(19:13) UK AISI

(19:56) Warning Shots Fired

(21:09) Well Did You Lead Him On? What Were You Wearing?

(25:05) Some People Have Principles

(26:58) Cause You're Living In (At Least) One

(27:51) What Happens Now?

(31:30) Oh How The Vibe Vibers Have Vibed

(33:36) We Now Know We Can Sometimes Do Things At Least?

(37:48) The Lighter Side

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

First things first: Claude Fable 5 is the new best publicly available model.

I have noticed a step change, where Fable can suddenly help me in ways that previous models were not worth bothering to query. Almost everything it has noticed in one of my drafts so far has been spot on and it is downright scary. Suddenly I am motivated to once again continue improving my Chrome extension. I only ask for things I actually want or am curious about, and it has nailed every question I have asked it.

That does not mean it is the right tool for every job.

There are four good reasons to often not use Fable.

---

Outline:

(02:05) Another Week Another Giant System Card

(03:02) How To Tell A Fable

(08:33) Why They Did That In That Way

(10:14) Why They Really Really Shouldn't Have Done That In That Way

(12:02) They Get Letters

(16:11) What's In A Name

(18:13) Executive Summary Of Their Executive Summary

(19:28) Introduction (1)

(19:55) RSP Evaluations (2.1 and 2.2)

(23:01) AI Research And Development (2.3)

(25:48) Alignment Risk (2.4)

(27:21) Cyber (3)

(30:30) Jailbreak Robustness

(32:04) Yay UK AISI

(32:32) Mundane Safety (4)

(34:26) Agentic Safety (5)

(36:19) Alignment (6)

(42:25) In Vendbench

(45:19) White Box Investigations (6.4)

(47:53) Grading Awareness

(51:20) Guess The Teacher's Password

(52:33) It Knows This Is A Test And This Is Fine

(56:03) I'm The Real Shady

(58:06) The Lighter Side

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

A lot happened this week, including a great trip out to Lighthaven.

The main event, the one that matters, was the release of Claude Fable 5. The public now has its hands on a Mythos-class model, alongside strong safeguards.

As always with a new model, I take a few days to draw in reactions, try out the model and read the system card, before I offer my takes, other than to say this is an extremely strong model. Full coverage of Mythos begins tomorrow with the model card, which will include discussion of the controversy over model safeguards.

This post is instead about all the things that did not involve Claude Fable.

Due to the time crunch from Claude Fable, I am also postponing my coverage of Dario Amodei's new essay, Policy on the AI Exponential, which I have not yet read.

Table of Contents

---

Outline:

(01:00) Language Models Offer Mundane Utility

(01:15) Language Models Don't Offer Mundane Utility

(02:31) Huh, Upgrades

(03:00) On Your Marks

(07:37) Choose Your Fighter

(10:56) Get My Agent On The Line

(11:14) Copyright Confrontation

(12:14) Serious Trouble

(13:01) Cyber Lack of Security

(13:21) A Young Lady's Illustrated Primer

(14:34) They Took Our Jobs

(17:48) The Art of the Jailbreak

(18:08) Get Involved

(21:54) In Other AI News

(23:02) Hand Over The Money

(24:37) Show Me the Money

(27:50) Quiet Speculations

(28:50) Quickly, There's No Time

(38:37) Super Secret Evals

(40:47) The Quest for Sane Regulations

(45:15) New Draft Bill Who Dis

(47:07) Slow Down There Good Buddy

(48:58) Chip City

(49:14) The Week in Audio

(49:54) People Just Say Things

(50:43) People Really Hate AI

(51:42) Rhetorical Innovation

(54:50) Aligning a Smarter Than Human Intelligence is Difficult

(56:15) Everyone Is Confused About Consciousness

(56:54) Cooperative Alignment

(01:02:23) Let Claude Chat

(01:04:31) The Lighter Side

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

The big story today is the release of Claude Fable 5, the version of Claude Mythos that Anthropic believes they can safely distribute to the people. You should absolutely be switching over to that model and trying it out. But as always, this blog does not rush into commenting on a new model until we have a few days to play around with it and see what our new baby can (and can’t) do. This will be no exception, and coverage of Fable in earnest will start Friday or Monday.

Today I instead bring you several related stories around policies and plans for AI, that came out before the Fable announcement.

First we have the Administration giving us an AI memorandum, that I read as an attempt to legally implement ‘Anthropic is fired forever and we will use any models we have for whatever we want no matter what’ combined with some good government and diffusion plans.

Second, OpenAI has come out with a plan for how to ensure AGI benefits everyone. It includes a very strong call for international coordination among key actors to ensure the ability to slow down AI development in the name [...]

---

Outline:

(02:29) An AI Memorandum

(09:56) Greetings From The Department of War

(10:52) Lab With a Plan

(17:20) A Difference Of Perspectives

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

---

Outline:

(04:16) What Do We Want?

(07:25) A National Framework

(11:18) Building State Capacity And CAISI

(12:31) Whole-Of-Government Resilience

(13:20) Reasonableness Rising

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.