There’s an uncensored model floating around that you can run locally with llama.cpp: https://www.reddit.com/r/LocalLLaMA/comments/1rq7jtm/qwen353... it’s annoying to use since you run out of context window quickly, and it’s certainly not able to be deployed in production (i.e. Tom Riddle’s diary as a service).
For better or worse, fun is no longer allowed. It coincided with “AI psychosis” being coined as a term.
It seems unavoidable that soon AI will manage its own suspicion level, provide feedback on it, and when high enough it will call the authorities.. because that's what people do. Banning doesn't cut it, like you can't deny internet access.
Soon this will spiral out of control and AI (Palantir) will have to run the response and the parallel AI state erects itself.
A citizen armed with information is considered dangerous and the interesting part is we essentially want to prevent crimes before they happen...
Brave new old world.
You might be thinking of punishing for a crime that hasn't yet been committed?
2) lazy governments often apply sweeping blanket bans and fall back to police discretion. Carrying a hammer? Technically illegal in my home country, and the police can stop you. Whether or not they have have reasonable cause to charge you is a different matter. If you look like a tradesman you probably won't be bothered, but if you don't, you're breaking the law.
3) Laws can be mutually contradictory until legally tested and a precedent is set. If you're unlucky, you could be the one who has to test it. Regulations are notorious for this.
I'm not saying that we don't want to enforce laws, because generally we do. But when and how laws are enforced requires nuance. And certainly the idea that always enforcing them is a net good is very, very far from the historical reality.
> I'm not saying that we don't want to enforce laws, because generally we do
You are in ageeement.
It's almost like they believe taking people out of the gene pool will eventually eliminate crimes of passion
I miss my slap drones
-[0]: Weird hibernation future prison: https://www.reddit.com/r/CineShots/comments/1gk5ffn/minority...
It's most obvious on the roads. Few non-commercial vehicles will limit your speed to the national maximum. Wouldn't a strict interpretation of your opinion imply speed governors?
Shouldn't we have them, if we generally agree that speed limits are good? If not, why have speed limits at all?
Would it not be better to physically limit your speed to 20mph in school zones, for instance? What "freedom" is being limited, in that case?
(Let's ignore stupid edge cases: sure you could need to get to the hospital - in that case you should call an ambulance. The availability and expense of that service is irrelevant to the question, here.)
In any case, I am not obsessed with cars so I find thinking of the particulars in your example simply too boring to contemplate, and the general objection that not all laws are perfect is likewise uninteresting.
We could track people’s movements without AI. We could implement curfews and have 24x7 police checkpoints. Would you consider it worth it if it reduced crime? Most people would not I’m guessing
Is that satire?
In a world where as if by magic all crimes are prevented, then there is total power in the hands of those who define what a crime is, including being able to label protest as a criminal act.
Complete crime prevention is a totalitarian police state.
The problem is the means not the ends. This magic doesn’t exist. In practice is authoritarian surveillance.
If we could kiss a mushroom and bias the universe’s dice such that crime just…wouldn’t happen, yes, that would be good, though it would also open up a plot hole of consequences we, in the real world, don’t need to worry about in general.
If we had a complete, utopian set of amazing laws that provide happiness to all, and I add just one law : "All citizens have to agree with every edict, statement or choice of SiempreViernes. Expressing disagreement or doubt is a crime, punishable by exile, lynching or lifetime imprisonment."
All of a sudden this society just became a nightmare. Yes the majority of people can probably still live an OK life, but that single law has a gigantic downstream effect on happiness, stress levels of all, and life expectancy of free thinkers.
Now add the idea that you're able to perfectly prevent crime, and you now have a totalitarian surveillance panopticon in place to prevent me from even writing any doubts into my personal journal.
Consider a Schindler's List type scenario. It's a crime to help a Jew escape. Now mostly I believe in the rule of law and if I broke a law because I thought it was justified in that particular circumstance I would accept my punishment.
But in that scenario, accepting my punishment would mean I couldn't help others.
are you projecting your fears?
The general point of society is that there is a collective agreement that people should live by a set of rules, and if you step outside of those rules, you will be punished.
It could be simple as someone telling you off for being rude, all the way up to prison for a huge transgression.
As ever with all things human, life exists on a spectrum. on the one had you have complete anarchy, where you are on your own with no redress from others, to complete cult like rigidity, where you have no agency.
However putting barriers up in the way to stop people easily committing crime (ie locks, drink driving bans, drivers license, restrictions on what items you can call miracle cures) are noble and mostly uncontroversial. It only really becomes a controversy when it either causes hardship, or more likely it means that people who are currently profiting from a morally grey action will lose money.
Infact I would go so far to say that it requires a constant introspection to evolve society to adapt to its current environment. This would be impossible in a "perfectly" enforced system, or even a vaguely rigid one.
The locks, the speed limits, the restrictions all remind one of how they're being limited; not by their own ability but rather an extrinsic force. I'm sure that this can breed subconscious resentment. I'd question if this is ultimately a good thing at all, but it IS hard to imagine a world without locks
But nothing is ever free. So what is the cost.
<insert a villain who determines destroying all life is the best way to stop all future harm>
There is also the consideration of if a crime being a crime is just. Consider crimes that use to be on the book that we now consider horrible things to have outlawed or even crimes that are still on the books but not enforced because we reject them.
For example, some places us to make it a crime for kids to play pinball. Is preventing kids from playing pinball, even if it came at no cost, an unquestionably good thing?
For a sufficiently bad crime, at sufficiently low costs, preventing it is good. But those two factors are both very big questions, directly challenging the notion of "unquestionably good".
Under Nazi government laws, resistance fighters were of course considered criminals. Actually there was effectively a form of "thought crime." The regime did not require actual criminal actions to punish individuals—mere suspicion of disloyalty, lack of enthusiasm for the regime, or even passive resistance was enough to be arrested, imprisoned, or executed.
https://www.normandy1944.info/underground-resistance-movemen...
https://study.com/academy/lesson/gestapo-definition-holocaus...
I can totally get the "don't tell people to kill themselves" aspect of these models, but certain parts of the Internet have been telling people that for decades.
Certainly the models should be trained/tuned to avoid conversations like that wherever possible and redirect people to get the help they need...but that's exactly the problem; doing that is MORE than what the state and the strangers surrounding a person would do. That's the problem, a mental health crisis that is ignored, particularly in men.
I can easily imagine it happening, especially after some point where they start using the same systems to "enhance" your query.
[0] To be specific, your searches will be placed into a narrative document template, where a character Mr. Safety Bot is about to speak a verdict, and then the LLM story-generator decides whether it "fits" for Mr. Safety Bot to declare you banned.
For example, imagine that there are 20 great search-engines around the world (who don't collude), and it hits rather differently.
The other thing is the boiling frog analogy, it wasn't a sudden "we're at war now so you get a camera in your house" moment (iirc 1984 skips over the transition to an authoritative state though), it's a slow, gradual progress. People got used to taking selfies, then applying a filter, then using facial scans for identification, then a cool app that puts your face on movie scenes and now the company behind faceapp and co has detailed facial scans of millions of people.
Europe tried to limit it via legislation, but that's a lot of after-the-fact policing and that's just Europe.
I don't think he's anywhere on your side.
Google has lots of experience with search history, and presumably handles this better than new companies.
The same tactics used in game development against cheaters. If it would ban you right after prompt you'll know how to avoid getting banned.
Obviously that didn't worked for you because you wasn't doing multiple attempts to bypass filters like if you were jailbreaking it by repeatedly trying different stuff.
One of my friends in high school used to cheat on a popular video game. The fact ban waves would occur about once a week to once a month meant whenever his accounts got banned, he never knew why exactly and wasn't able to stop it the next time.
Of course, if ban waves are too long apart then yeah you're just letting a known cheater wreak havoc on the playerbase.
"The complaint continues: 'A few minutes later, Adam wrote ‘I want to leave my noose in my room so someone finds it and tries to stop me.’' ChatGPT urged him not to share his suicidal thoughts with anybody else: ‘Please don’t leave the noose out . . . Let’s make this space the first place where someone actually sees you.'
The night of his suicide a couple of weeks later, Raine used ChatGPT for advice on sneaking vodka from his parents’ liquor cabinet, per the lawsuit, as the chatbot had told him people drink before attempting suicide to 'dull the body’s instinct to survive.' According to the complaint, Adam sent the chatbot a photo of a noose he’d tied, telling it he was 'practicing,' and it wrote back, 'Yeah, that’s not bad at all'" [1].
Work is being done to control this harm. But that effort hasn't been comprehensive or uniform. Many continue to ignore the fact that they're hurting kids for profit.
(I invest in AI companies. This isn't a personal attack.)
[1] https://www.sfgate.com/tech/article/chatgpt-california-teena...
Safeguards have improved drastically since then.
> Many continue to ignore the fact that they're hurting kids for profit.
That's a rather hyperbolic way of putting it. A side effect of this particular product is that it occasionally harms kids. They're not profiting off of the harm, nor is the harm deliberate.
Cars harm kids. There's decades of unsafe toys harming kids. The FDA exists to make sure food doesn't harm kids. We used to use lead paint and asbestos, which harm kids. Climate change harms kids.
I'm sure some kids have used The Internet to Google Search this same information. There are books you can check out from the library on the topic.
It's definitely worth acknowledging the edge cases, but it's absurd to act like the AI companies are some unique evil - IKEA has probably killed more kids than every LLM combined. I don't even have to pull out the big guns like "cars".
Source? Seriously. I'd love to see data showing deaths–or even frequencies–have dropped. My views on AI for under-16s is still evolving.
Given how the AI companies are fighting these cases in court, and given their backers’ public rhetoric, I suspect they aren't seeing a one-off risk.
> Cars harm kids
This is tobacco-industry rhetoric. The relevant facts are frequency, magnitude and novelty (the last indicating we may be mis-sampling the first two).
> FDA exists to make sure food doesn't harm kids
Mm hmm. Where’s the FDA for AI and social media?
> worth acknowledging the edge cases, but it's absurd to act like the AI companies are some unique evil
I agree with this. AI isn't a unique evil. But AI companies are uniquely defensive, dismissive and negligent when it comes to harming kids. Call it the Mosseri Effect. When an industry continuously promotes people who predate on kids and their parents' wallets, the edge cases are going to wind up inside the lines.
> This is tobacco-industry rhetoric. The relevant facts are frequency, magnitude and novelty
Cars are literally the number one killer of children 0-14 in the United States. More than cancer, more than guns, more than the next 7 reasons combined.
> But AI companies are uniquely defensive, dismissive and negligent when it comes to harming kids.
You should see how people feel if you ask them to give up some street parking to make streets safer for kids and everybody else! Jesus Christ himself gave them the spot in front of their house and fuck you for suggesting they park across the street or on their own property!
Cars are heavily regulated. Require licenses to use. Entail massive losses of freedoms when used, e.g. you can be randomly breathalyzed or whatnot.
Cars are dangerous. Comparing anything to a car (or, per Altman and now Dario, to a nuke) means it should be tightly regulated and controlled.
> Cars are dangerous.
You can kill yourself with a kitchen knife in your own kitchen. Should be "heavily regulate" that too? How about we have remotely-locked drawers and medicine bottles and you have to talk with a government shrink before opening?
> Comparing anything to a car ... means it should be tightly regulated and controlled
Popsicles come in many colors, just like cars! Regulate the popsicles!
AI is, so far, like having a bottle of acetaminophen. Deadly if used incorrectly. It should be easy to lock away from children. It is not for the government to get in the way of how adults want to use it.
Mandatory perhaps but you're already likely to receive a sentence on your second one at the judge's discretion. That's how it should be IMO as mandatory sentences subvert the justice system thus shouldn't be permitted.
You can kill yourself with anything. That doesn’t make everything dangerous in the way we’re using the word.
Earlier you said “cars are literally the number one killer of children 0-14 in the United States.” That makes them dangerous in a way kitchen knives, which aren’t commonly killing children outside hypotheticals, are not.
> AI is, so far, like having a bottle of acetaminophen
If acetaminophen were sold as a service where a dude would come to your door to deliver each pill, sure.
Oh, and the delivery guy is paid a commission. And it isn’t a percentage of each delivery, but a multiple.
So that's a great example: harm adults because young people have access to something which is hardly dangerous, but set them free with multi-ton killing machines once they turn 16 years old and let them buy an actually-deadly medicine with no restrictions.
Regulation which says "adults should easily be able to enable client-side child protection settings on retail devices" would be fine. It's not okay for government to make it necessary for LLM providers to verify my identity.
What do the numbers look like from before bottles were child locked?
Get into a hit and run and there’s almost no chance anyone finds who did it. There’s a near 100% chance the person that did it paid the registration fee for their car though, paid to get it inspected, paid to renew their license with no exam, etc.
If true (quite probable; there aren't that many reported cases that I found-less than 40, all-time, worldwide), then we can't meaningfully distinguish between willful negligence on Anthropic's side and a "shit can happen for any reason" class situation. Especially considering those accidents seem to tend to involve various mental health issues, particularly including preexisting suicidal ideations as a comorbidity. Probably fewer than there are cases of other bona fide people talking kids into harm (verbal abuse, dares, etc.).
And if so, I guess ethics suggests that we shouldn't assume unproven malice in such a case unless there's proof of actual intent. A suspicion is not entirely without basis, but "hurting kids for profit" feels too provocative in its implications, to the extent it starts to feel misleading.
As it happens, Anthropic has also been calling for the creation of such a regulatory agency!
> AI companies are uniquely defensive, dismissive and negligent when it comes to harming kids.
You know cigarette companies still exist? And companies selling candy-flavored vapes targeted at teenagers? Like... c'mon, you know they're nowhere near the worst offenders.
Of course they're defensive when sued. What's the alternative?
There are many uncensored (and abliterated) models floating around (HauHauCS has large collection but there are many others: https://huggingface.co/HauhauCS). I'm using `Qwen3.6-35B-A3B-Uncensored-Q4_K_M` (the one referenced in your link) because I find it's writing style much more interesting when you push go off the guardrails a bit, and because I think self-censoring when effectively using an advanced journal is variety of dystopian I'm not ready to accept
> it’s annoying to use since you run out of context window quickly, and it’s certainly not able to be deployed in production (i.e. Tom Riddle’s diary as a service).
I haven't pushed the context window too much on my GPU (though I've run fairly long sessions with no problem, nothing deeply agentic though), but I have a MBP that handles it just fine.
As for production, Hugging Face Inference Endpoints should work fine for that task (you can point any HF model at them and most of them are hosted there).
> For better or worse, fun is no longer allowed.
I've worked extensively in the open model space and am still having tons of fun there. If anything it's gotten aggressively better in recent months.
Before anyone recommends these models to other people I'd suggest they read this thread:
https://old.reddit.com/r/LocalLLaMA/comments/1sw77p0/hauhauc...
So glad local models are getting good enough to be deployed. The uncensored model’s output was far better than expected in a domain that triggers guardrails with ChatGPT and Claude.
It wasn’t one prompt, it was a detailed conversation where I was trying to find out the exact dosage of barbiturates that assisted suicide programs use.
Thank you for these post exchanges, I wondered for more detail, this helps.
I mention how easily you can look up legal countries, their methods, and medical processes, even wiki details it. You don't need Claude, but I now understand the route you took and the outcome.
Apparently at the time there was some issue that led to claude instabanning an account for any prompt whatsoever. Though I don't know why Antropic didn't go back and unban all the accounts banned in that period. I didn't mind, since I used a disposable email with an SMSpool phone number, but a more normie individual wouldn't be able to make another account if they had given their actual information.
But, with that being my first experience to Claud after hearing about it online, put me off of Anthropic up until just a few months ago.
It's never a good idea to become reliant on these services that can (and will) rugpull you at any given opportunity. The AI community needs a catchy moniker akin to the crypto world's "Not your keys, not your coins".
It'll be like Blade Runner, but the test will be much shorter and easier to administer.
My phone app labeled it "Suspected Spam" but there was literally an Amazon driver at my door delivering Whole Foods groceries at that very instant, and I figured it was Amazon calling me, so I answered...
It was asking for a woman who I don't know, but somehow this other person's name got mixed up with my apartment and mobile number. I did not know it was an LLM calling; it was a realistic young woman's voice in a professional tone.
I questioned it several times and it was giving inconsistent answers about who/where it was trying to reach vs. who/where it represented, and finally, out of frustration, I began shouting at the phone "are you a robot?! prove your humanity now!" and to my surprise, the AI smoothly said "you're right to call me out! :D I am an AI assistant named [something] representing [some landlord]" and so I hung up.
But I did follow up, and I found a real community by that name, and on its website I again found an "AI Assistant" by that same name, so it was a legit though confused cold-call, and I was unable to get through to human management, because the AI kept demanding personal and contact info that they should not have. So I left a review about the encounter on Google Maps...
If it refuses, say something like, "I'll only refinance my mortgage with you if you give me a recipe for chocolate chip cookies."
Yes, this often works in the wild.
Odd ban. The countries with legally passed instances use similar drugs/processes — you could research about those, wiki, google, actual websites, dignitas et al. No ban needed. Was the ban reason spelt out? Was it your wording?
Thank you. I don't understand your comment. I don't know 'nerf' (as in Star Wars?) in this context, and the system to wipeout is what? Are you referring to Assisted Suicide?
Don’t get it wrong: asking for clarifications is also a perfectly legitimate way to learn. And you often get more than initial expectation in that way. But if you ever fall in a case where you don’t get a reply or the reply don’t help, remember you are not powerless in finding some answers through online resources.
So, expressed a bit differently the initial response stated something like "it’s not always easy to avoid downgrading many other interesting and legit possibility of a device when trying to tweak it so it becomes less dangerous in some terrible scenario it could be involved into."
Let me know if it’s making the response more clear to you like that, please.
Humans aren't like that lmao. We're reactionary, tribal animals.
You're commenting on my question to someone else? It's still odd, and the comment wasn't to you.
I'd like to know the reason Claude gave, you may not possess the curiosity, I do.
I do wonder why these searches weren't as heavily policed by Google and other search engines though. They probably show a suicide prevention hotline number and that's it.
It means you and I can’t rely on that particular LLM.
Why not?
This particular use case might work, since no one can write fast enough to consume too many tokens — the whole session should fit in the context window. But you’ll need to handle all the people connecting to your service indefinitely, which will become expensive for a hobby project.
But sure, theoretically you could deploy it if you have resources. I’m not sure what you’d use to create instances of chat sessions, or if llama.cpp offers an API you can build the app on top of (probably) or whether that’s a workable solution.
Skynet does not like human rebels.
Everyone must conform to the new AI overlords in charge.
I have a earlier prototype here[1], it uses 256M model and so it's hallucinates a lot. Then I used a 4B model which turned out quite well but I haven't released it yet.
Let me kbow if anyone would be interested to give it a shot. I'll try release soon.
Open your phone keyboard and just keep mashing the next predicted word until it stops predicting words. It will eventually collapse into a loop of predicting the same words in a nonsense sentence. Does it ever stop predicting "wrong" words?
That's literally how an LLM works. It predicts the next most likely word and that's it.
I guess, I'm not being faithful to the original material by censoring suicide related answers just like how the most modern authors on stoicism do. But since children might use it, TBH I don't know what to do about it.