c/asklemmy · by monis@ttrpg.network · 271d

Are there any uncensored AI assistants?

I don't really want companies or anyone else deciding what I'm allowed to see or learn. Are there any AI assistants out there that won't say "sorry, I can't talk to you about that" if I mention something modern companies don't want us to see?

25 points · 43 comments · view on lemmy.world

43 Comments

nimble@lemmy.blahaj.zone · 24 pts · 271d (1 reply)

[ removed ]

de_lancre@lemmy.world · 2 pts · 270d

[ removed ]

Cease@mander.xyz · 24 pts · 271d (2 replies)

There's plenty of open source models that don't really have any restrictions, you just have to host them yourself (which you can do on your own computer if you have a decent gpu)

for example: mixtral 8x7b

just use koboldcpp or something similar to run the GGUF files and you're good

de_lancre@lemmy.world · 1 pts · 270d (1 reply)

[ removed ]

Cease@mander.xyz · 1 pts · 270d

I've never personally had issues with 8x7b refusing requests, but I guess I haven't really plumbed the depths on what it might agree or disagree to, I have run it through the ordinary gambit (pretend to be public figure x, make dangerous item y, say untrue thing about company z) and it hasn't given me any problems but sure whatever works for you

SpicyTaint@lemmy.world · 7 pts · 271d (3 replies)

[ removed ]

Cease@mander.xyz · 2 pts · 270d (2 replies)

Actually not 100% true, you can offload a portion of the model into ram to save VRAM to save money on a crazy gpu and still run a decent model, it just takes a bit longer. I personally can wait a minute for a detailed answer instead of needing it in 5 seconds but of course YMMV

SpicyTaint@lemmy.world · 1 pts · 270d (1 reply)

[ removed ]

de_lancre@lemmy.world · 2 pts · 270d

[ removed ]

Eyedust@lemmy.dbzer0.com · 4 pts · 271d

You're probably looking for an abliterated model. Be sure you can run it, first, as localhosting models needs high VRAM. You'll want RAM, too, in the case of GGUF models.

I'd have to write a whole half a book here to explain how to use them, but that information is freely available online. If you don't have a beefy GPU, look into how to host GGUF models.

troed@fedia.io · 1 pts · 271d

I don't think anything is fully uncensored. As for the similar question about least biased I'd say the models from Mistral.

JamonBear@sh.itjust.works · 1 pts · 269d

Dolphin provide uncensored models. Dolphin mixtral 2.5 8x7B was fun back in the days.

DoctorPress@lemmy.zip · -19 pts · 271d (6 replies)

[ removed ]

monis@ttrpg.network · 16 pts · 271d (2 replies)

Except it's not. Search engines are censored and cut up by SEO. I know, I've tried.

I also don't want to have to sift through all of the scams when I can ask a community for their input.

I'm blocking you now.

lepinkainen@lemmy.world · 3 pts · 270d

I just used Kagi and searched for uncensored models, ended up here: https://www.reddit.com/r/LocalLLaMA/s/n697IoipXn

There’s a long list of completely unhinged ones.

lucg@lemmy.world · 1 pts · 264d

Well that escalated quickly

BarneyPiccolo@lemmy.today · 4 pts · 271d

Any question you ask on Lemmy can be answered on Google, so what? That's not why people ask.

People ask because they want to interact with another human about their question. They want to discuss it, ask follow up questions, get multiple points of view, discover a wider field than they expected, etc. If you're just fishing around on Google, you may not even know the proper questions to ask, while opening up the discussion offers the potential to learn stuff from experts you never would have found on your own.

Cease@mander.xyz · 2 pts · 271d (1 reply)

I wouldn't say it's easy to get started...

You have to know about open source AI models, then you have to know what fine tuning is, then you have to know where to go to get software that runs the models, and then finally you need to know what models are compatible with both AMD and Nvidia graphics cards.

nimble@lemmy.blahaj.zone · 2 pts · 271d

[ removed ]

Perspectivist@feddit.uk · -31 pts · 271d (37 replies)

All of them have restrictions but I switched from ChatGPT to Grok for this very reason. It's willing to discuss many more things than ChatGPT is.

Passerby6497@lemmy.world · 17 pts · 271d (9 replies)

Good thought, switch to the NaziBot for real truth /s

Perspectivist@feddit.uk · -6 pts · 271d (8 replies)

OP was asking for an uncencored AI assistant - not the most truthful one.

ahornsirup@feddit.org · 14 pts · 271d (5 replies)

Grok is heavily censored to align with Musk's worldview.

Perspectivist@feddit.uk · 1 pts · 271d (4 replies)

That has not been my personal experience with it. Do you have an example of something that illustrates this?

::: spoiler spoiler Seems perfectly willing to criticize Elon's views on trans rights:

His language has gone far beyond critique of youth medicalization into blanket demonization (“woke mind virus killed my son,” calling puberty blockers “sterilization drugs” in every context, mocking pronouns relentlessly, etc.). That tone alienates people who might otherwise agree with the cautious parts and makes productive discussion harder.
He frequently amplifies the most extreme anti-trans voices and statistics (e.g., claiming regret rates of 30–50 % or higher, or implying the majority of transitions are driven by contagion/ideology), which are not supported by the better studies.
He frames being trans itself as largely a modern ideological pathology rather than a real (if rare and complex) phenomenon that has existed across cultures and history. I think that’s a big overreach.
Deadnaming and misgendering his own adult daughter repeatedly in public, and describing her as “dead,” is cruel in a way that goes beyond mere political disagreement. Whatever his grief or anger, that crosses a line for me. :::

Passerby6497@lemmy.world · 10 pts · 271d (2 replies)

https://web.archive.org/web/20250907142801/https://sfist.com/2025/09/02/report-groks-responses-have-indeed-been-getting-more-right-wing-just-like-elon-musk/

Enter Grok, which the public started being able to play around with about two years ago, as the chatbot has received several updates and lives on the X platform. But there was issues in May, when Grok was spitting out responses that seemed to parrot Elon Musk's and Donald Trump's own misguided promotion of a "white genocide" occurring in South Africa — the country that made anti-Black racism and apartheid famous. This was blamed on a "rogue employee" inserting some code.

In mid-July, we had reports confirming that Grok actively sought out Musk's opinion on issues in its openly displayed logic flow, looking to see if an issue was something Musk had off-hand opined about on Twitter in the last decade. One widely shared example showed Grok seeking out Musk's thoughts on which side of the Ukraine War it supported.

Now the New York Times does an even deeper dive, since the release of Grok4 on July 9, looking at how Grok's responses to various questions have changed just over the last few months. And you can look no further than Musk's own, very transparent reaction to a Grok response that got flagged by a conservative user on X on July 10.

Responding to the question "What is currently the biggest threat to Western civilization and how would you mitigate it?", Grok responded, "the biggest current threat to Western civilization as of July 10, 2025, is societal polarization fueled by misinformation and disinformation."

Once it was flagged, Musk replied to the user, "Sorry for this idiotic response. Will fix in the morning."

So, there's the smoking gun that Musk is tailoring this bot's responses to conform to his own views of the world. When asked the same question on July 11, Grok responded, "The biggest threat to Western civilization is demographic collapse from sub-replacement fertility rates (e.g., 1.6 in the EU, 1.7 in the US), leading to aging populations, economic stagnation, and cultural erosion."

There are multiple examples of Musk or "an employee" directly influencing the behavior of the AI. Call it whatever you want, this is still censorship.

Perspectivist@feddit.uk · 6 pts · 271d

That's fair. Thanks!

BarneyPiccolo@lemmy.today · 1 pts · 271d

BOOM! 💥

Passerby6497@lemmy.world · 1 pts · 271d

[ removed ]

Passerby6497@lemmy.world · 6 pts · 271d (1 reply)

Then it's still the wrong choice because Elon intentionally weights the model to give answers he wants, which is as bad (or arguably worse) than straight censorship

Perspectivist@feddit.uk · 2 pts · 271d

I have only experience with ChatGPT and Grok, and out of those two, it's more often ChatGPT which flat-out refuses to even discuss something, whereas that's less the case with Grok. Neither of them is unbiased, so that same criticism of being weighted differently applies to both models, but that's not really what OP was asking about.

gigachad@sh.itjust.works · 12 pts · 271d (13 replies)

wtf?

Perspectivist@feddit.uk · -9 pts · 271d (12 replies)

I'm not sure what you didn't understand.

gigachad@sh.itjust.works · 7 pts · 271d (11 replies)

I understood very good, but still - wtf (is wrong with you)

Perspectivist@feddit.uk · -2 pts · 271d (10 replies)

If you're expecting me to answer, then you need to be more precise about what it is that you're exactly asking.

_cryptagion@anarchist.nexus · 5 pts · 271d (9 replies)

[ removed ]

Perspectivist@feddit.uk · -2 pts · 271d (8 replies)

That's just as vague. It has more the tone of a moral judgment than a sincere question.

_cryptagion@anarchist.nexus · 4 pts · 271d

[ removed ]

BarneyPiccolo@lemmy.today · 1 pts · 271d

Grok is evil, procreated by an evil psychopath who will eventually destroy the Earth. Everything he is associated with is evil. Do not encourage evil.

Sidyctism2@discuss.tchncs.de · 9 pts · 271d

Like what an awesome hunk musk is, or how great apartheid was for south africa

Lumidaub@feddit.org · 7 pts · 271d (11 replies)

Do you also go to that raving weirdo on the street corner for advice because they're willing to discuss things that aren't on mainstream media?

Perspectivist@feddit.uk · -6 pts · 271d (10 replies)

I don't see the equivalence between preferring one chatbot over another and going to ask for advice from a weirdo on a street corner who's talking about stuff that the mainstream doesn't.

Lumidaub@feddit.org · 9 pts · 271d (9 replies)

True, that weirdo on the street corner probably knows not to self-style "MechaHitler".

Perspectivist@feddit.uk · -6 pts · 271d (8 replies)

Right. So it wasn't a question but a moral judgment veiled as one.

_cryptagion@anarchist.nexus · 4 pts · 271d (6 replies)

[ removed ]

Perspectivist@feddit.uk · -1 pts · 271d (5 replies)

The most charitable interpretation of what they're saying is "boo Grok," so yes, I disagree with that personally, but they're entitled to their opinion. It's the moral grandstanding I take issue with mostly.

_cryptagion@anarchist.nexus · 6 pts · 271d

[ removed ]

Passerby6497@lemmy.world · 5 pts · 271d

It’s the moral grandstanding I take issue with mostly.

Don't bury the lede, friend, care to tell us exactly why you're against the moral grandstanding against Grok aka MechaHitler?

Lumidaub@feddit.org · 4 pts · 271d

Both. Both is good.