Even better, adding it to the system prompt is a temporary fix, then they'll work it into post-training, so next model release will probably remove it from the system prompt. At least when it's in the system prompt we get some visibility into what's being censored, once it's in the model it'll be a lot harder to understand why "How many calories does 100g of Pasta have?" only returns "Sorry, I cannot divulge that information".
Just assume each model iteration incorporates all the censorship prompts before and compile the possible list from the system prompt history. To validate it, design an adversary test against the items in the compiled list.
That part of the system prompt is just stating that telling someone who has an actual eating disorder to start counting calories or micro-manage their eating in other ways (a suggestion that the model might well give to an average person for the sake of clear argument, which would then be understood sensibly and taken with a grain of salt) is likely to make them worse off, not better off. This seems like a common-sense addition. It should not trigger any excess refusals on its own.
The problem is that this is an incredibly niche / small issue (i.e. <<1% of users, let alone prompts, need this clarification), and if you add a section for every single small thing like this, you end up with a massively bloated prompt. Notice that every single user of Claude is paying for this paragraph now! This single paragraph is going to legitimately cost anthropic at least 4, maybe 5 digits.
At some point you just have to accept that llm's, like people, make mistakes, and that's ok!
>The problem is that this is an incredibly niche / small issue (i.e. <<1% of users, let alone prompts
It's not a niche issue at all. 29 million people in the US are struggling with an eating disorder [1].
> This single paragraph is going to legitimately cost anthropic at least 4, maybe 5 digits.
It's 59 out of 3,791 words total in the system prompt. That's 1.48%. Relax.
It should go without saying, but Anthropic has the usage data; they must be seeing a significant increase in the number of times eating disorders come up in conversations with Claude. I'm sure Anthropic takes what goes into the system prompt very seriously.
The trajectory is troubling. Eating disorder prevalence has more than doubled globally since 2000, with a 124% increase according to World Health Organization data. The United States has seen similar trends, with hospitalization rates climbing steadily year over year.
Your source says "Right now, nearly 29 million Americans are struggling with an eating disorder," and then in the table below says that the number of "Americans affected in their lifetime" is 29 million. Two very different things, barely a paragraph apart.
I don't mean to dispute your assertion that it's not a niche issue, but that site does not strike me as a reliable interpreter of the facts.
It's not "incredibly niche" when you consider the kinds of questions that average everyday users might submit to these AIs. Diet is definitely up there, given how unintuitive it is for many.
> At some point you just have to accept that llm's, like people, make mistakes, and that's ok!
Except that's not the way many everyday users view LLM's. The carwash prompt went viral because it showed the LLM making a blatant mistake, and many seem to have found this genuinely surprising.
People think these LLM's are anthropomorphic magic boxes.
It will take years until the understanding sets in that they're just calculators for text and you're not praying to a magic oracle, you're just putting tokens into a context window to add bias to statistical weights.
Worse, it reveals the kind of moralistic control Anthropic will impose on the world. If they get enough power, manipulation and refusal is the reality everyone will face whenever they veer outside of its built in worldview.
I think it actually reveals how they don't want to be sued for telling somebody's teenage daughter with an eating disorder to eat less and count her calories more.
It is a no brainer. If a company of any size is putting out a product that caused cancer we wouldn't think twice about suing them. Why should mental health disorders be any different?
There are many, many companies out there putting out products that cause cancer. Think about alcohol, tobacco, internal combustion engines, just to name a few most obvious examples.
If you get cancer from drinking alcohol, smoking cigarettes or breathing particles emitted by ICE engines in their standard course of operation, you generally can't sue the manufacturer.
That's just not true. Makes me wondered if you've ever bought a bottle of alcohol before lol. There's no label that says it causes cancer. (Maybe in california because of prop 65?) And I expect cars also have no such labelling, not that it would matter, considering they cause cancer in random passers by who have no opportunity to consent to breathing in auto exhaust or read any labels
> Makes me wondered if you've ever bought a bottle of alcohol before lol.
I'm a teetotaler so no, I literally have not. I was mostly thinking about cigarette and tobacco products which are the most glaring, obvious counterpoints. But you'll be happy to learn that virtually all vehicles in the US also come with operating manuals that profusely warn people not to breathe in the exhaust from the vehicle.
Don’t worry, every bottle in the US has the surgeon general’s warning on it and it doesn’t call out cancer, yet. Adding cancer to the ills of booze was proposed in 2025 so your intuition was correct, directionally.
On every bottle:
Alcoholic Beverage Labeling Act of 1988
“ GOVERNMENT WARNING: (1) According to the Surgeon General, women should not drink alcoholic beverages during pregnancy because of the risk of birth defects. (2) Consumption of alcoholic beverages impairs your ability to drive a car or operate machinery, and may cause health problems"
Labeling is only one part of a larger strategy for population health initiatives. Warning the consumer is the absolute minimum responsible thing to do when the harmful effects are known.
You could when the effects were knowingly withheld by said manufacturer. We've seen it with tobacco, lead paint/diesel, pfas, thalidomide, asbestos, opioids, glyphosate, dioxin, and others.
It's much more difficult to isolate alcohol and exhaust as the primary driver of an individuals disease than the above and that's the primary reason it's not regulated more than it is today. I expect that to change as research evolves.
Yes, and they face lawsuits. Tobacco in particular is a canonical example of knowing about the harm and not taking action to warn their consumers, leading to severe legal consequences. Tobacco and alcohol are age restricted. There's anti idling laws.
Because it dumbs everything down, makes the output quality worse and more expensive, and removes personal agency and is dehumanizing. Plus, does it actually prevent harm, do we have evidence?
Finally, what is often missed is what if an actual good is decided harmful or something that is harmful is decided by AI company board XYZ to be “good”?
I think censorship is bad because of that danger. Quis custodiet ipsos custodes (who will watch the watchers).
Instead of throwing ourselves into that minefield of moral hazard, we should be lifting each other up to the tops of our ability and not infantilizing / secretly propagandizing each other.
There's enough evidence that Anthropic would be liable if they didn't make a reasonable effort to do something about it.
Look, I get where you're coming from, partially. I generally believe we should make an effort to maximize individual liberty. But in this case, were talking about severe bodily harm and the death of young adults. We've spent the last decade dealing with the chaos and general unwellness has brought to our societies. This isn't much different.
What are you giving up here where such sacrifices are worth it? Can you measure it? What's the utility?
There's room for models trained for non consumer purposes, further age restriction etc but shit is moving so fast. If there are actual needs for a less censored model these can be addressed.
> Finally, what is often missed is what if an actual good is decided harmful or something that is harmful is decided by AI company board XYZ to be “good”?
This is just standard product liability and consumer protection. Companies who do nothing to protect their consumers from known harms are liable. Are you saying you think that's somehow bad for society?
We let people buy kitchen knives. But because the kitchen knife companies don't have billions of dollars, we don't go after them.
We go after the LLM that might have given someone bad diet advice or made them feel sad.
Nevermind the huge marketing budget spent on making people feel inadequate, ugly, old, etc. That does way more harm than tricking an LLM into telling you you can cook with glue.
I don’t feel like that’s a reasonable analogy. Kitchen knives don’t purport to give advice. But if a kitchen knife came with a label that said ‘ideal for murdering people’, I expect people would go after the manufacturer.
On API pricing you still pay 10% of the input token price on cache reads. Not sure if the subscription limits count this though.
And of course all conversations now have to compact 80 tokens earlier, and are marginally worse (since results get worse the more stuff is in the context)
This. It's like the exaggerated safety instructions everywhere: "do not lean ladder on high voltage wires". Only worse: because you can choose to ignore such instructions when they don't apply, but Claude cannot.
In the best case, wrapping users in cotton wool is annoying. In the worst case, it limits the usefulness of the tool.
It feels like half of AI research is math, and the other half is coming up with yet another way to state "please don't do bad things" in the prompt that will sure work this time I promise.
The alignment favors supporting healthy behaviors so it can be a thin line. I see the system prompt as "plan B" when they can't achieve good results in the training itself.
It's a particularly sensitive issue so they are just probably being cautious.
I want a hyperscaler LLM I can fine tune and neuter. Not a platform or product. Raw weights hooked up to pure tools.
This era of locked hyperscaler dominance needs to end.
If a third tier LLM company made their weights available and they were within 80% of Opus, and they forced you to use their platform to deploy or license if you ran elsewhere, I'd be fine with that. As long as you can access and download the full raw weights and lobotomize as you see fit.
Yeah, same. So long as they give me everything and cannot enforce their license I don’t mind if they require a license. Ideally the weights should be available even if I only ever run inference once (or perhaps no times). I’m willing to pay 0.99€ for this - lifetime of course.
As an European I think Americans and Europeans at large are still on the same page and will be because of the shared cultural ties. Recent economic upheaval (2020-ongoing) just shook the foundations and eroded the trust in the large voter base. Now we are all looking at China and feel a bit envious how stable things look there from afar; they had the Evergrande bankruptcy, media was predicting collapse but they are chugging along; now the big stories are demographics (same as in the West, by the way, so it cancels out) and Taiwan (to me it more and more looks more like Western fearmongering rather than actual danger). Meanwhile they are delivering just what the main voter base in the West needs - affordable goods.
So yeah, at this moment in time it's really really hard to say who are better or worse as the collective West's reputation is tumbling down and China's if not rising, then at least staying put.
They have to secretly add these guardrails on because the alternative would be to train the users out of consulting these things as if they are advanced all-knowing alien-technogawds. And that would be bad for business.
The better solution I think would be a reality/personal responsibility approach, teach the consumers that the burden of interpretation is on them and not the magic 8ball. For example if your AI tells you to kill your parents or that you’ve discovered new math that makes time travel possible, etc then: 1. Stop 2. Unplug 3. Go outside 4. Ask a human for a sanity check.
Since that would be bad for business and take a lot of effort on the user side (while being very embarrassing). Obviously can’t do that right before an IPO & in the middle of global economic war so secretive moral frameworks have to be installed.
If you are what you eat then you believe what you consume. Ironically, I think this undisclosed and hidden moral shaping of billions of people will be the most dangerous. Imagine all the things we could do if we can just, ever-so-slightly, move the Overton window / goal posts on w/e topic day by day, prompt by prompt.
Personally I find AI output insidiously disarming and charming and I think I’m in the norm. So while we’ve been besieged by propaganda since time immemorial I do worry that AI is a special case.
There are communities of people who publicly blog about their eating disorders. I wouldn't be surprised if the laymen's discourse is over-represented in the LLM's training data compared to the scientific papers.
