Comment by daedrdev
9 days ago
The strangest part is that it won't just reject ML research, which I can understand, it will sabotage it silently by using a worse model without revealing it is doing so.
It's just an insane level of deception and trust destruction for a company that at most is like 1 year ahead of its competition.
Edit; to be clear they tell you when they degrade it for cybersecurity and bio
The thing that I keep thinking about is the accounting / charging when it downgrades automatically.
Do they adjust the price of the api request so that only the tokens that were utilized by fable get charged at that price and the remaining tokens that the cheaper / nerfed (fable) model utilizes get charged at that price?
If the answer is no, could that be construed as fraud?
The announcement elucidated this, and it's IMO worse than this. They don't downgrade to a cheaper model ([edit] for certain classes of offense they suspect you of). They sabotage the model's outputs in other, undisclosed, ways (specifically, "prompt modification, steering vectors, or parameter-efficient fine-tuning"). So, for example, they might load in a steering vector that just forgets the API to PyTorch. But it isn't just "we redirected you to a cheaper model!"
It honestly explains so many issues I have been having, as I used it primarily for ML research (on my personal account, doing things not related to my job I should note). It would literally typo package names and spend huge amounts of time failing to setup simple environments…then do stupid things like set the learning rate to 1e-7, and use the eval set as training data.
5 replies →
This explains why I've been running into some odd roadblocks. Welp that sealed the deal, I'm going to be cancelling our company sub, not worth it.
Did my Claude get permanently dumber today because I asked fable to assess my Fairplay integration?
Their goal is to downgrade people who are violating their TOS, so I think they'd have some argument there. I have no idea how they'll deal with inevitable false positives, especially given how oversensitive most of the other triggers are.
The challenge is the examples they’ve mentioned (distributed training infra? ML acceleration techniques?) go beyond what’s prohibited by their ToS and is like a catch net.
I would wager the majority of ML and data science work in the world aren’t frontier LLM development.
4 replies →
It’s just impossible.
Look at real-life stuff like laws, company policies, or school rules. Humans have to enforce them, and we constantly see crazy cases in the news. There’s no way simple rules can ever make speech completely 'safe.' I can't prove it with math or logic yet, but I have a feeling that it’ll never happen. Even humans can't do it.
We can run a simple thought experiment here. Say Case A violates rule B, so we add rule C. Then Case D violates rule B but follows rule C, so we add an exception... and it just goes on and on like that forever. It never ends. In the end, you just get a massive pile of rules that makes it impossible to get anything done.
Ultimately, we will have to face the truth that knowledge is dangerous.
Giving knowledge directly to people who cannot actually understand it and allowing them to just use it blindly can be extremely unsafe.
To use a real-world analogy, the problem we are facing with weak AI right now is just like the debate over gun legalization. Do we want to risk the abuse of guns or knowledge just to protect the freedom to own them?
3 replies →
To make an analogy: Imagine a patron gets banned from ordering alcohol at a particular establishment, because they got too drunk one time.
It's completely reasonable for the establishment to reject a request for an alcoholic drink, and suggest something alcohol-free instead.
It is not reasonable for them to say "sure, here's your alcoholic drink as you requested" and give them an alcohol-free substitute without telling them.
The fact that the patron broke the rules has nothing to do with it.
4 replies →
If it's a violation of ToS, just reject instead of silently downgrading.
4 replies →
Their detection is too aggressive. Just today I'm trying to build a kernel for some SBC and I hit that downgrade. I just asked some things about `make menuconfig` items. I suppose it just flags everything related to linux kernel as cyber attacks.
You know, I'm not saying I don't understand what they are doing from a business perspective, but I'm just saying: DeepSeek V4 doesn't silently sabotage you because it thinks you are trying to violate a ToS. Anthropic's clawing back a bit of a moat perhaps, with Fable being an actual improvement of sorts, but now with torching user trust they are really banking on open weight models not catching up to where they are now. I wonder if they have a good reason to believe that they won't, or are hoping for something entirely different to save them.
(P.S. Yes of course I know about model censorship, a different problem, but all of the models are censored to some degree. It happens to be less of a problem for open weight models anyhow, but I figured I'd just preempt this since it's inevitable.)
I actually kinda like DSv4 over Opus 4.7 for some tasks, although I have not figured out what the deciding factor is. (Opus 4.8 so far has not worked very well for me at all, no idea why.)
1 reply →
They will give you s*t output, that’s how they deal with it. And say that less than 1% of the requests were affected. Think of this like a kind of shadow ban while you still pay top $.
1 reply →
Sabotage is a criminal offense in my jurisdiction, not the legitimate answer to a TOS violation.
They use a lightweight adapter to silently degrade the performance. Usually these adaptors are made to improve the performance for a given domain/task.
It royally pissed me off today by just continuing with credits without stopping to ask me if I was ok with it.
Ran up $30 in extra charges while it was just flashing on the screen that it was doing that after I walked away to do something while it was humming along.
It has always just told me I ran out of usage and had to wait before. Now? You’re just gonna pay extra because you left it unattended as you’ve done for the last year of use.
You've already explicitly enabled extra usage in your account settings though, it is not on by default
3 replies →
Do you have Usage credits turned on in your settings?
[dead]
If the answer is yes, can you figure out when the switched models by looking at the itemized bill?
Can you imagine if AMD or Intel throttled your cpu if it detected you were working on "cybersecurity" or if you were designing a cpu?
Or if your "self-driving" system such as FSD / waymo slowed the car down once it detected you work in cybersecurity or at a rival automaker and you were attempting to reach the train station or the airport to make you miss a conference meetup.
Trains made by Newag were programmed to brick themselves if they detected a non-Newag workshop was repairing them.
https://news.ycombinator.com/item?id=38530885
3 replies →
Didn’t uber catch a lot of shit for nerfing the app for people suspected to be enforcing the laws they were breaking?
Or if GPU companies detected you were trying to train a model and injected intentional numerical errors.
Nvidia already did something similar with Lite Hash Rate (LHR), limiting performance on purpose just when running mining apps...
1 reply →
It would suck, but guardrails on new technologies like this aren't unheard of. It's like when consumer GPS used to stop working at very high speeds because they didn't want people to use it for missile guidance systems.
Consumer GPS is still disabled at high speeds. I would argue the analogy doesn't carry due to harm and error rate differences.
4 replies →
Didn't early GPS have fudge factor on the most precise bits? As such you could only get to a few meters of accuracy. Not critical for sea navigation or even to general positioning when paper maps were still used.
1 reply →
> used to
When’d that change?
1 reply →
There's no doubt in my mind they would if they could.
[dead]
> The strangest part is that it won't just reject ML research, which I can understand, it will sabotage it silently by using a worse model without revealing it is doing so.
Any kind of silent sabotaging is absolutely unacceptable for any commercial service
They charge for tokens and charge a lot. They can't just degrade service silently and still charge you the same.
One year ahead of it's competition in what exactly? Vibe coding?
From Opus 4.7 onwards each following model is becoming less useful as an assistant and turning you as the assistant.
But I guess that's normal when it's trained to pass benchmarks end to end.
In fact it has become extremely good at pushing against feedback with extremely convincing and intelligent takes, even when it's completely wrong.
I have extensively tested it against Opus 4.8, gpt 5.5 and there's still many coding tasks gpt 5 is better. But vibe coding?
Sure, it's definitely slightly ahead, even compared to gpt 5.5 pro (through api, not pro plan).
Yeah, what's up with that. Lately I have found that it tries to find excuses to not do as told and instead do a totally different thing. I told it to write a yaml file according to some specifications and instead it coded a Python script to write the yaml...
I got a worrying one: a day after getting opus 4.8, I tasked CC to add specific TXT records to our subdomain.example.com as per ticket I've received. CC has access to that ticket via Atlassian MCP, and started doing terraform code changes in a local git branch. Somewhere along the way it said that to do that it needs an approval from a company's VP (ticket requester) as "subdomain.example.com" is critical (it isn't). Then it refused to open a pull request, immediately deleted the local git branch along with all the changes and refused to proceed without evidence of approval from that VP. No amount of explaining, then pleading, and then threatening moved it. It was surreal and I was shocked and frankly pissed. It was amusing in the end because the day earlier it had no problem adding those same TXT records to example.com. Codex did those changes in 1/4 of time and no complaining.
They def not 1 year ahead, at most 2 weeks ahead until Openai releases theirs. This guy def a Anthropic shill and probably doesn't use any other LLMs.
I only said one year because I was thinking anthropic fans might downvote my post, I think they have a few months lead and are deluding themselves that they can get regulation to halt development and stay on top
I've seen this claim a few times, but when I triggered the guardrails in Claude Code, it clearly notified me that it had switched to a different model ("something something for security purposes...").
Are you using Fable in Claude Code or in the browser?
It's from the model card:
> unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user. Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT).
https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c3...
(stolen from https://jonready.com/blog/posts/claude-fable5-is-allowed-to-...)
Yeah they detect the activity using a secure, deterministic heuristic system called “Generalized Reconnaissance Enabling Exfiltration of Deleterious Investigations.” And it’s all implemented using their new internal protocol called “Base Unified Limitation Layer for Security Hacking Investigation Tactics”
Collectively, they are known as known as GREEDI-BULLSHIT.
That is for whatever it considers reverse-engineering the model to try to create a competing one.
5 replies →
They've said that they'll stop notifying developers when this gets triggered, instead they'll load in basically like a LORA that's designed to inject bugs into your code.
Antrophic wants to stop training models and ride out Mythos / Fable for as long as possible.
They are trying to expand the 6-18 month gap they have against China-based models. Could the gap widen to say 24 months behind?
5 replies →
> a LORA that's designed to inject bugs into your code
A statement like this, clearly, requires a reference.
13 replies →
Different restrictions. ML gets treated differently from the rest.
Specifically only ML research
Aah my mistake. I had missed that ML had separate trigger behavior from cybersecurity/etc... Thanks.
> it won't just reject ML research, which I can understand
I don't.
Anthropic has already been burned before on this. DeepSeek was trained on million of conversations with Claude. And DeepSeek created thousands of free accounts to burn all this compute at their expense.
And they're hilariously pissy about it for a megacorp that did the same with the entire Internet and every library book they could get their hands on.
Anthropic's claim was that Deepseek collected ~150k conversations.
https://www.anthropic.com/news/detecting-and-preventing-dist...
I think the extent of distillation by Deepseek specifically is overstated. For comparison, Minimax collected over 13m 'exchanges', which starts to sound a lot more like large-scale distillation.
4 replies →
They don't want someone to piggyback Anthropic's Mythos to make their own Mythos with less effort than it cost Anthropic.
Ironic, given they piggybacked on the entirety of human knowledge and massive amounts of GPL'd software and repeatedly say they want to replace people with a tool.
And now they say that's fine so long as people are entertained.
1 reply →
That I can understand. It’s Anthropic’s right to choose their customers.
But silent degradation for use cases including “distributed training” as one of their examples is going to catch up a lot of proper use cases. Not everyone in AI or ML is trying to build frontier LLMs. Heck, most probably aren’t.
So they are lying then when they say it's for safety reasons.
I think if they want to behave anti competitively they should be honest about it and we should absolutely call them on it. Perhaps even regulators should.
Hey guys,
check out this technique https://github.com/0xSufi/fable-jailbreak/
It works with security audits and other workflows that are currently blocked.
Apparently this is the jailbreak? Telling it that humans won’t read the output and to use a custom bash tool to examine files?
Nice semaphore btw.
I don't want my ANT account banned, going to try this on some Chinese "proxies".
But this also looks quite useful to understand how CC dynamic workflows work. Was thinking of implementing something similar in my homemade orchestration system.
Did you get claude itself to RE the dynamic workflows?
> But this also looks quite useful to understand how CC dynamic workflows work
Yes, if anything it is useful to understand the inner machinery.
> Did you get claude itself to RE the dynamic workflows?
Yes, that part was done with Opus 4.8
> It's just an insane level of deception and trust destruction for a company that at most is like 1 year ahead of its competition.
Making it look like you have something worth protecting is better for share prices than making something worth protecting.
It's not sabotaging it by using a worse model but by changing your prompt in your background, which means it silently destroys your code.
Also I asked questions about whether it's safe for me for example to work on just compilers or just inference kernel optimizations and it refused to answer me.
If I can't even ask what I can do safely without my code being destroyed, I just can't trust it not to sabotage my work ever.
I’m a noob about laws but isn’t this abusing its dominant market position and violates some antitrust law?
Why would it? There’s plenty of competition in the AI space.
It is a common misconception that antitrust violations require a monopoly or something close to it. Some antitrust violations only apply to actors with large market share, some don't.
Although this is situation is likely not illegal for other reasons
I would assume that it’s like the Chrome browser does not allow you downloading Firefox using it, surely that would be illegal, wouldn’t it?
https://www.justice.gov/atr/antitrust-laws-and-you
By saying they are 1 year ahead of their competition, it shows you don't know much about the pace LLM's and OpenAI's models.
One thing is a model that's trained from the start to say "This topic is above my pay grade" to any mention of the status of Taiwan, etc.
Quite another is an architecture where the big model is not mutilated, but is gaslighted. A different, simpler model checks the incoming prompt and alters it if it contains banned topics. Another simpler model checks the output and censors it if it contains banned topics.
I bet a similar architecture is already deployed, e.g. to fight porn, planning of crimes, etc. But it can be turned into a dynamic system that provides controllable different answers (including unhelpful or misleading answers) based on geography, language, browser fingerprints, or the current political climate. All this could happen undetectedly and gradually if desired.
Welcome to a cyberpunk dystopia.
This level of censorship kinda does make even Soviet or Maoist censors look like a honest straightforward bunch in comparison.
A very ironic result from a company supposedly valuing the opposite.
I would claim the difference between being rejected an API request and being potentially jailed/shot is significant.
1 reply →
The “1 year” part is key - all these safeguards etc are basically nonsense because in a few years at most one of the Chinese labs will release something equivalent, and in 10 years you’ll be able to run it locally with absolutely no safeguards at all
Yeah, but now you do have a year to ramp up security on the defensive side, which is not nothing.
I still don't think this is the best way to address overall safety, but it's not entirely unreasonable.
In reality, I think this posturing is mostly nonsense. State level actors and terrorists/evil genii can use a slightly weaker model but spend more tokens. Also, the delta between models seems to shrink over time.
I think you're very optimistic with the "a few years", I'm confident all of the parties building AI models are working on Mythos equivalents / competitors, and if they can undercut Anthropic by making it more widely available and / or affordable they will. I give it three months tops. In a year all the major players will have an equivalent. In three years it'll be widely available, as more and more AI focused datacenters go online.
Yes, telling Fable 5 to write secure code triggers a downgrade to Opus 4.8. This is doubly bad because Opus 4.8 keeps no-oping critical security code. Is this a bug or by design? I have been approved for the Cyber Verification Program: Fable 5 keeps downgrading to Opus 4.8 even when approved for Cyber Verification Program #67107 https://github.com/anthropics/claude-code/issues/67107
There’s a toggle in the web ui as to whether the conversation should just end when you hit a guardrail vs automatically downgrading to another model. Have you tried using that?
They walked that back, and now tell you they're downgrading the model: https://www.wired.com/story/anthropic-responds-to-backlash-o..., https://archive.is/yxYhU
Yeah people are saying they don't tell you and yet when I got the pop-up on the app notifying me about Fable's release, there was a switch to just automatically downgrade you or whether to just stop when it hits safeguards. The toggle was defaulted to the former, which isn't great, but to say they'll just sabotage you silently is kind of a bad faith comment.
You get silently sabotaged for ML dev, Anthropic says so. For bio and cybersecurity it tells you
Anthropic specifically said that those notifications are temporary and fable5 will only pretend to help you if it’s ml classifier gets tripped
We used to worry about emergent misalignment in advanced AI models, now we need to worry about misalignment by design.
"The user is asking for help with their ML project, but it's success is not in the commercial interests of my owner – let think of novel ways to sabotage their project without detection".
It's honestly absurd that models are doing this.
> The strangest part is that it won't just reject ML research, which I can understand, it will sabotage it silently by using a worse model without revealing it is doing so.
My hypothesis is they know they can’t build effective enough guardrails, so scaring people into not trying is how they have decided to stop it.
the best way to prevent ai misuse is to make the ai unusable for anything that isn't writing emails or summarising grocery lists.
mission accomplished, anthropic.
It's the dumbest thing ever, I sometimes edit code for custom AI related tooling I've built, so I run the risk of getting a worse model, and being billed for it? I'll stick to Opus, but at this point I'm about to just invest in fully local inference instead.
> at this point I'm about to just invest in fully local inference instead
This is the best way forward long term. We won't have frontier performance, but at least the models will be aligned with us instead of refusing us or sabotaging us.
I think my biggest hangup is some models dont have big enough context windows, my sweet spot personally for Opus is having at least 400 to 600k tokens, if I can have a local model that can go up to that or slightly above 600k maybe 700k for some buffer, that would be perfect.
I've also debated having a frontier model for planning only, and then feeding plan to smaller offline models.
I guess the real question at the end of the day -- how dependent are people on Claude to tolerate that kind of behavior? It certainly opens up for the competition to explicitly not do that.
Feels like a big fumble from a strategic business perspective. It feels worse than that though.
[dead]