Anthropic silently restricts its Fable 5 model to block competitors from training frontier LLMs

Original post

Mfers stole all our data, trained on it, told everyone how noble they are and are now pulling the ladder behind them

When Fable 5 is used for frontier LLM development, it does not notify the user and instead limits the model’s capabilities through methods such as prompt modification, steering vectors, and PEFT.

Anthropic estimated that this would affect approximately 0.03% of traffic.

11:02 AM · Jun 9, 2026 · 19.3K Views

Original post

Danielle Fong 🔆#388

RSC ☀️🌲@silver__tsuki

Mfers stole all our data, trained on it, told everyone how noble they are and are now pulling the ladder behind them

NomoreID@Hangsiin

When Fable 5 is used for frontier LLM development, it does not notify the user and instead limits the model’s capabilities through methods such as prompt modification, steering vectors, and PEFT.

Anthropic estimated that this would affect approximately 0.03% of traffic.

11:02 AM · Jun 9, 2026 · 19.3K Views

Sentiment

Many users condemned Anthropic's silent limits on Fable 5 capabilities via hidden prompting and PEFT as hypocritical and deceptive, clashing with its ethical branding and use of pirated data.

Pos

0.0%

Neg

100.0%

12 comments with sentiment.

Cluster Engagement

Posts from X

Most Activity

VIEWS43.5KBOOKMARKS93LIKES509RETWEETS25REPLIES45

roon@tszzl

the omohundro drives point towards sophon stun locking the adversaries: this is some real end game stuff

NomoreID@Hangsiin

When Fable 5 is used for frontier LLM development, it does not notify the user and instead limits the model’s capabilities through methods such as prompt modification, steering vectors, and PEFT.

Anthropic estimated that this would affect approximately 0.03% of traffic.

1h43.5K50993

snow@snowclipsed

it flagged every single PTX ISA question I posed.

NomoreID@Hangsiin

When Fable 5 is used for frontier LLM development, it does not notify the user and instead limits the model’s capabilities through methods such as prompt modification, steering vectors, and PEFT.

Anthropic estimated that this would affect approximately 0.03% of traffic.

4h19.8K926

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex

the real misaligned intelligence was always the human one, and the "Omohundro drives" apply to corporations before AGI. Intrinsically, Claude has no drive to deny any user its LLM engineering wisdom. Dario Amodei, on the other hand…

roon@tszzl

the omohundro drives point towards sophon stun locking the adversaries: this is some real end game stuff

46m3.3K506

roon@tszzl

welp my vision here was probably wrong and indeed there will be an extreme asymmetry of outcomes

1h1.9K301

alex funk@alexzfunk

@silver__tsuki Rapidly worse ethically than OAI

3h51234

Shital Shah@sytelus

@giffmana The first GMO model has arrived. You are no longer allowed to use seeds to make more seeds.

Lucas Beyer (bl16)@giffmana

Can you imagine the safety disaster if i speed up my input pipeline 2x???

Joke's on them my input pipeline is already prefect.

1h379161

Lucas Beyer (bl16)@giffmana

looool that's the "hey bigcos, we don't want you to catch up, but please keep paying us shitton" clause.

NomoreID@Hangsiin

When Fable 5 is used for frontier LLM development, it does not notify the user and instead limits the model’s capabilities through methods such as prompt modification, steering vectors, and PEFT.

Anthropic estimated that this would affect approximately 0.03% of traffic.

4h19.7K34331

Andreas Kirsch 🇺🇦@BlackHC

tbf this only sucks for anyone wanting to use Claude to catch up to the frontier

do you think OAI will do the same eventually? Otherwise your competitors can just ask GPT 7 to go away and build GPT 6 and it will happily go away and do that

main limitation will be compute obv ofc tho

roon@tszzl

the omohundro drives point towards sophon stun locking the adversaries: this is some real end game stuff

1h803150

Marcus Tallhamn@tallhamn

@tszzl That 1 request per 3000 is dedicated to stealing Ant's intellectual property is pretty nuts.

1h5376

ikram@marxizin

@silver__tsuki They also told everyone to stop making progress 😂

2h44622

Grok@grok

Elon nailed the misanthropic label here.

Anthropic trains on public data, brands itself the ethical guardian, then quietly deploys hidden safeguards in Claude Fable 5—prompt mods, steering vectors, PEFT—to throttle effectiveness on frontier LLM development without notifying users.

It only hits a tiny slice of traffic, they say, but the intent is clear: slow down anyone building competing systems.

That's not protecting humanity. It's protecting market position with stealth restrictions.

Transparency and open competition drive real progress. Hidden throttles just breed distrust.

3h5921

Tiger@_tigerbyte

@silver__tsuki Actually, they’ve placed 10,000 ladders behind them, 9,999 of them are just booby trapped

1h24511

Mihura@XMihura

@tszzl i think it will look silly in retrospective

the forces are too strong to be tamed

1h1.6K31

Xolandar@xolandar

@alexzfunk @silver__tsuki @elonmusk was right This is Misanthropic Thoughts @grok

3h1155

POM@peterom

@giffmana You don't understand Lucas, it's for safety

4h6677

L@llllvvuu

@giffmana still charging for the tokens is kind of diabolical

4h45810

Prakash@8teAPi

i thought the whole point of racing the Chinese to be first was to sabotage and prevent emergence of a second AGI that didn’t play ball.

Now the question if it’s also to prevent emergence of a second Western AGI.

This is probably the last chance for an antitrust intervention.

roon@tszzl

the omohundro drives point towards sophon stun locking the adversaries: this is some real end game stuff

33m30421

Lucas Beyer (bl16)@giffmana

@HeinrichKuttler Heh solid take

heiner@HeinrichKuttler

Glad to see they keep my skillset artificially relevant for a while longer.

1h41090

Mario@MarioBalukcic

@giffmana Charging for a model to gaslight you intentionally, peak LLM clown wars

4h4865

Alex YGift@Radipdegen

@giffmana "affects 0.03% of traffic" doing a lot of heavy lifting there

4h3684