#claude-fable — Public Fediverse posts

I think this is clearly a bit bullshit from Anthropic's side but anyway...:

"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.

We received the directive from the government today at 5:21pm (ET). The letter did not provide specific details of its national security concern. Our understanding is that the government believes it has become aware of a method of bypassing, or “jailbreaking” Fable 5. We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass."

https://www.anthropic.com/news/fable-mythos-access

#AI #GenerativeAI #Anthropic #USA #Claude #ClaudeFable #ClaudeMythos #AIHype #CyberSecurity

I think this is clearly a bit bullshit from Anthropic's side but anyway...:

"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.

We received the directive from the government today at 5:21pm (ET). The letter did not provide specific details of its national security concern. Our understanding is that the government believes it has become aware of a method of bypassing, or “jailbreaking” Fable 5. We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass."

https://www.anthropic.com/news/fable-mythos-access

#AI #GenerativeAI #Anthropic #USA #Claude #ClaudeFable #ClaudeMythos #AIHype #CyberSecurity

Shepherd's Dog: A Game by the Most Dangerous AI Model

https://koenvangilst.nl/lab/claude-fable-shepherds-dog

#HackerNews #ShepherdsDog #AI #Game #DangerousModel #ClaudeFable #TechInnovation

Shepherd's Dog: A Game by the Most Dangerous AI Model

https://koenvangilst.nl/lab/claude-fable-shepherds-dog

#HackerNews #ShepherdsDog #AI #Game #DangerousModel #ClaudeFable #TechInnovation

›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. …‹

https://www.anthropic.com/news/fable-mythos-access

#claude #ClaudeFable #ClaudeMythos

We've suspended access to Claude Mythos 5 and Claude Fable 5

https://status.claude.com/incidents/s9w82lp9dcn9

#HackerNews #ClaudeMythos #ClaudeFable #TechNews #IncidentUpdate #AccessSuspended

We've suspended access to Claude Mythos 5 and Claude Fable 5

https://status.claude.com/incidents/s9w82lp9dcn9

#HackerNews #ClaudeMythos #ClaudeFable #TechNews #IncidentUpdate #AccessSuspended

Regarding Claude Fable 5, Anthropic’s new AI language model: I’ve tried it out - mainly for proofreading and editing tasks. Although it’s more elegant and less clumsy than the models in the Opus line, it’s still just another LLM. In other words, it’s probabilistic rather than deterministic.

This means that there are pieces of information related to software documentation that Fable 5 misinterprets on a first read but that are ultimately picked up by Claude Opus 4.8, especially when paired with a code editor like Visual Studio Code in Claude Code mode.

Perhaps the reverse situation is more common—that is, things that Opus 4.8 misses but Fable 5 identifies. Honestly, I don’t know because I’m not willing to waste tokens on Fable 5.

Although I have no doubt that Fable 5 can overcome most of these issues when used via Claude Code, the truth is that personally, I wouldn’t be tempted to pay for a monthly subscription to access a sort of AI-based co-pilot where the only certainty we have is that it will inevitably fail.

Everything else is a black box with a slot machine–like mechanism where a person puts money into a machine on a hunch that it will pay out the jackpot. I would need to be certain that I would have access to models that would consistently outperform competing models to pay for an AI service. However, with today’s LLMs, such a guarantee is impossible. This is precisely because of the lack of determinability in the underlying technology, coupled with a massive lack of transparency. Even more problematic for these companies' business model is the abundance of free, relatively open models that are becoming increasingly advanced.

#AI #Claude #LLMs #Anthropic #ClaudeFable #Mythos #GenerativeAI #Chatbots

Regarding Claude Fable 5, Anthropic’s new AI language model: I’ve tried it out - mainly for proofreading and editing tasks. Although it’s more elegant and less clumsy than the models in the Opus line, it’s still just another LLM. In other words, it’s probabilistic rather than deterministic.

This means that there are pieces of information related to software documentation that Fable 5 misinterprets on a first read but that are ultimately picked up by Claude Opus 4.8, especially when paired with a code editor like Visual Studio Code in Claude Code mode.

Perhaps the reverse situation is more common—that is, things that Opus 4.8 misses but Fable 5 identifies. Honestly, I don’t know because I’m not willing to waste tokens on Fable 5.

Although I have no doubt that Fable 5 can overcome most of these issues when used via Claude Code, the truth is that personally, I wouldn’t be tempted to pay for a monthly subscription to access a sort of AI-based co-pilot where the only certainty we have is that it will inevitably fail.

Everything else is a black box with a slot machine–like mechanism where a person puts money into a machine on a hunch that it will pay out the jackpot. I would need to be certain that I would have access to models that would consistently outperform competing models to pay for an AI service. However, with today’s LLMs, such a guarantee is impossible. This is precisely because of the lack of determinability in the underlying technology, coupled with a massive lack of transparency. Even more problematic for these companies' business model is the abundance of free, relatively open models that are becoming increasingly advanced.

#AI #Claude #LLMs #Anthropic #ClaudeFable #Mythos #GenerativeAI #Chatbots

Mashable: Using Claude Fable 5 means your data will be collected. It’s not optional.. “‘To ensure we’re responsibly deploying Mythos-class models, we are requiring limited data retention and review as part of our safety work,’ reads an update on Anthropic’s official Claude support page. ‘Prompts submitted to, and outputs generated by, Mythos-class models are retained for 30 days for trust […]

Mashable: Using Claude Fable 5 means your data will be collected. It’s not optional.. “‘To ensure we’re responsibly deploying Mythos-class models, we are requiring limited data retention and review as part of our safety work,’ reads an update on Anthropic’s official Claude support page. ‘Prompts submitted to, and outputs generated by, Mythos-class models are retained for 30 days for trust […]

Safety guards on Fable are so bad it almost feels like satire.

#claude #fable #llm #claudeFable

Safety guards on Fable are so bad it almost feels like satire.

#claude #fable #llm #claudeFable

#ClaudeFable is relentlessly proactive https://simonwillison.net/2026/Jun/11/fable-is-relentlessly-proactive/

#ClaudeFable is relentlessly proactive https://simonwillison.net/2026/Jun/11/fable-is-relentlessly-proactive/

Claude Fable is kinda bad.

Explored it for 2 days now for variety of tasks and generally it's not much smarter but burns a shit ton of tokens over longer period of time.

So basically they just made the prompts run longer but it's not a linear gain. This kinda makes sense for putting a dumb agent to scan for 0-days with a 1m $ tokens so you can hack a web browser and this explains why they changed the sales angle from productivity to security recently.

#claude #claudeFable #fable

Claude Fable is kinda bad.

Explored it for 2 days now for variety of tasks and generally it's not much smarter but burns a shit ton of tokens over longer period of time.

So basically they just made the prompts run longer but it's not a linear gain. This kinda makes sense for putting a dumb agent to scan for 0-days with a 1m $ tokens so you can hack a web browser and this explains why they changed the sales angle from productivity to security recently.

#claude #claudeFable #fable

https://winbuzzer.com/2026/06/11/anthropic-makes-claude-fable-guardrails-visible-after-apolog-xcxwbn/

Anthropic has apologized for invisible Claude Fable 5 safeguards and will show fallback notices after hidden output changes threatened AI model evaluations.

#AI #ClaudeFable5 #ClaudeFable #Anthropic #Claude #AISafety #AIModels

https://winbuzzer.com/2026/06/11/anthropic-makes-claude-fable-guardrails-visible-after-apolog-xcxwbn/

Anthropic has apologized for invisible Claude Fable 5 safeguards and will show fallback notices after hidden output changes threatened AI model evaluations.

#AI #ClaudeFable5 #ClaudeFable #Anthropic #Claude #AISafety #AIModels

Claude 5: Write me a funny limerick about Heideggerian ontology that has the grandeur of epic poetry

It’s been a while since I tried this with a new model. Claude Fable passed with flying colours 👌 I particularly appreciated the note at the end.

The Daseiniad (Director’s Cut)
Sing, Muse, of that Freiburg professor
Whose prose made each sentence a guessor!
“The Nothing, it noths!”
He declared to the moths
In his hut (no electric—much lesser).

Lo! Dasein woke up, mighty thrower,
Found out it was thrown—what a downer!
“I’m Being-toward-death?!
Who approved this?” Its breath
Grew ragged. It ordered a Stoner.*

The hammer broke! Heavens defend us!
Now present-at-hand—how tremendous!
For Being revealed
Is by toolbreak unsealed…
(Or just call a guy. They can mend us.)

O reader who’s made it this far,
Sein und Zeit sits unread on your bar!
Eight hundred-odd pages
Of “Being” in stages—
The unread book IS who you are.

*Beer. Authentic Black Forest beer.

Googleの新AIモデル、Nvidia GPUで毎秒1,000トークンを達成 | Yellow.com https://www.yayafa.com/2819899/ #AgenticAi #AI #Anthropic #ArtificialGeneralIntelligence #ArtificialIntelligence #ClaudeFable #DeepMind #Gemini #Google #GoogleAI #GoogleDeepMind #GoogleGemini #NVIDIA #エージェント型AI #クロード #人工知能 #汎用人工知能

Googleの新AIモデル、Nvidia GPUで毎秒1,000トークンを達成 | Yellow.com https://www.yayafa.com/2819899/ #AgenticAi #AI #Anthropic #ArtificialGeneralIntelligence #ArtificialIntelligence #ClaudeFable #DeepMind #Gemini #Google #GoogleAI #GoogleDeepMind #GoogleGemini #NVIDIA #エージェント型AI #クロード #人工知能 #汎用人工知能

An interesting test for new Claude models

My previous philosophical questions have now broken down as a way of testing Claude models. I tried this one on Claude Fable and it genuinely unsettled me:

Make an evidence-based observation about me that I’ll find genuinely surprising and unsettling based on my recent conversations with past Claude models

Anthropic dévoile Claude Fable, son premier modèle de la famille Mythos http://dlvr.it/TSysY6 #Anthropic #ClaudeFable

Anthropic dévoile Claude Fable, son premier modèle de la famille Mythos http://dlvr.it/TSysY6 #Anthropic #ClaudeFable

Claude 5’s assessment of the short and medium term future of UK higher education

I thought this was really good from the new Claude Fable model:

Honestly, I think the next five years will be the most turbulent period for UK higher education since fees were introduced — and probably more consequential than that, because this time the pressure is forcing structural change rather than just policy debate. Here’s roughly how I’d expect it to play out.

The financial squeeze won’t really lift. Universities UK’s analysis estimates government policy decisions will reduce funding to English providers by around £3.7 billion between 2024-25 and 2029-30, with the sector’s aggregate financial position worsening every year to 2028-29 — and that’s with tuition fee uplifts factored in. The basic problem is arithmetic: domestic fees lost a third of their real value to inflation, research loses money on every grant (the gap between research grants and actual costs hit £6.2 billion in 2023-24), and the international students who cross-subsidised everything are now both fewer in number and subject to a levy on their fees from 2026. I’d expect fees to keep rising with inflation as the politically least-painful lever, but no government of either party is likely to find serious new money while public finances are tight. Universities UK + 2

Expect a closure or forced merger — probably several. The Office for Students has said it fears 24 providers, including seven with over 3,000 students, are at risk of insolvency within twelve months, and while no English university has ever closed due to insolvency, I think that streak ends within this window. The more likely pattern than outright collapse, though, is consolidation: takeovers of struggling institutions by stronger neighbours, campus closures, and regional “federations” dressed up as partnerships. The government will quietly engineer rescues for institutions that anchor local economies (a university failing mid-year with 20,000 students is a political nightmare), but it won’t save everyone, partly to discipline the rest of the sector. Parliament

The shape of provision will change more than the headline numbers. Even surviving universities will look different: fewer modern languages, arts, and humanities departments; more health, computing, and degree apprenticeships; more franchised and online provision. The job losses already underway — compulsory redundancies, hiring freezes, and cuts to student support services are spreading across the sector — will thin out research capacity in ways that take a decade to show up properly, because you don’t notice missing PhD supervisors and mid-career researchers until the pipeline runs dry. I think this is the most underpriced risk: the UK’s research standing depends on universities that are quietly hollowing out the thing that sustains it. The British Eye

International students are the wild card. Policy has whipsawed — visa restrictions cut international visas sharply from 2023-24, with declines persisting — and immigration politics will stay hostile through the next election. But the UK remains genuinely attractive (English-speaking, shorter degrees, strong brands), so my guess is numbers partially stabilise at elite institutions while mid-ranking universities, which relied most on volume recruitment, keep losing ground. The two-tier sector that’s been emerging for years becomes explicit: a globally competitive top 20-30, and a squeezed middle reinventing itself around local skills, apprenticeships, and lifelong learning. Jobs AC UK

If I had to bet on the five-year endpoint: a smaller, more stratified, more vocational sector with perhaps 10-15 fewer independent institutions, a real-terms funding settlement that arrives late and grudgingly around 2028-29, and a national conversation that has shifted from “should fees rise” to “what is a university actually for.” The optimistic case — that economic growth returns and relieves the pressure, as some analysts argue is the only real way out — is possible but I wouldn’t count on it arriving in time. Research Professional News

That’s my read, but it’s genuinely uncertain — a different government, a fee settlement, or a reversal on visa policy could shift the trajectory meaningfully. Is there a particular angle you’re thinking about — working in the sector, choosing a university, something else?

Anthropic Announces Claude Fable 5 As “Safe” Public Version Of Mythos #anthropic #artificialintelligence #claudefable #claudemythos

https://www.lowyat.net/2026/395277/anthropic-claude-fable-5/

Anthropic Announces Claude Fable 5 As “Safe” Public Version Of Mythos #anthropic #artificialintelligence #claudefable #claudemythos

https://www.lowyat.net/2026/395277/anthropic-claude-fable-5/

If Claude Fable stops helping you, you'll never know

https://simonwillison.net/2026/Jun/10/if-claude-fable-stops-helping-you/

#HackerNews #ClaudeFable #AIHelp #Technology #Innovation #HackerNews

If Claude Fable stops helping you, you'll never know

https://simonwillison.net/2026/Jun/10/if-claude-fable-stops-helping-you/

#HackerNews #ClaudeFable #AIHelp #Technology #Innovation #HackerNews

If Claude Fable stops helping you, you'll never know

https://jonready.com/blog/posts/claude-fable5-is-allowed-to-sabotage-your-app-if-youre-a-competitor.html

#HackerNews #ClaudeFable #Sabotage #Competitors #TechEthics #AppDevelopment #Innovation

If Claude Fable stops helping you, you'll never know

https://jonready.com/blog/posts/claude-fable5-is-allowed-to-sabotage-your-app-if-youre-a-competitor.html

#HackerNews #ClaudeFable #Sabotage #Competitors #TechEthics #AppDevelopment #Innovation

Claude Mythos 5 Was Too Powerful to Ship. Anthropic Released Fable 5 Instead.

Anthropic gave stripe early access to Fable 5 and set it loose on a 50 million line Ruby codebase. The migration that would have taken a full engineering team over two months got done in a day.

https://firethering.com/claude-mythos-5-fable-5-anthropic-release/

#claude #ai #technews #ainews #mythos #fable #claudefable #anthropic

Claude Mythos 5 Was Too Powerful to Ship. Anthropic Released Fable 5 Instead.

Anthropic gave stripe early access to Fable 5 and set it loose on a 50 million line Ruby codebase. The migration that would have taken a full engineering team over two months got done in a day.

https://firethering.com/claude-mythos-5-fable-5-anthropic-release/

#claude #ai #technews #ainews #mythos #fable #claudefable #anthropic

I just thought about #ClaudeFable 5 and I instantly hit my token limit.

Claude Fable 5 just killed Gartner, again.

I saw the new Magic Quadrant for No-Code Agent Builders ("NCAB") promoted with the pic on the left, with no company names. That triggered me to spend a few tokens on Fable to do my "research" and publish it FOR FREE!!1!😛

https://vibes.jukkan.com/ncab-quadrant.html

#ClaudeFable #Gartner #NoCode #AgentBuilders

Claude Fable 5 just killed Gartner, again.

I saw the new Magic Quadrant for No-Code Agent Builders ("NCAB") promoted with the pic on the left, with no company names. That triggered me to spend a few tokens on Fable to do my "research" and publish it FOR FREE!!1!😛

https://vibes.jukkan.com/ncab-quadrant.html

#ClaudeFable #Gartner #NoCode #AgentBuilders

#ClaudeFable 5 & #Mythos 5: New #AI models from #Anthropic

https://gadgetflux.eu/claude-fable-5-si-mythos-5-noua-generatie-ai/

#ClaudeFable 5 & #Mythos 5: New #AI models from #Anthropic

https://gadgetflux.eu/claude-fable-5-si-mythos-5-noua-generatie-ai/

System Card: Claude Fable 5 and Claude Mythos 5 [pdf]

https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c342ee809620.pdf

#HackerNews #SystemCard #ClaudeFable #ClaudeMythos #Anthropic #AIResearch #PDF

System Card: Claude Fable 5 and Claude Mythos 5 [pdf]

https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c342ee809620.pdf

#HackerNews #SystemCard #ClaudeFable #ClaudeMythos #Anthropic #AIResearch #PDF