Discussion Does anyone here know an AI that is able to discuss suicide?

[Dingo23]

I've tried to discuss suicide with different Ai services. My problem is that every single one of them is censored when the topic revolves around methods or emotional preparation for the final moment. They usually just post help hotlines or outright stop the discussion by disconnecting.

Does anyone know an AI assistant that is actually uncensored in that regard? If not, that's ok.

 

[TheEndIsPeace]

Your best bet would be an uncensored local model.

 

[Aknu132]

Everytime an AI sends me a suicide hotline it triggers the fuck out of me, it shows how fucked up the world is, sending people to that shit service, no one truly wants to hear us.

 

[Cauliflour]

Why are you asking on the suicide forum of all places for AI chatbots to discuss suicide? Wouldn't it be more beneficial to ask somebody here? Not to mention AI is terrible for advice as it's a hardcore yes man that makes stuff up and says it with the uttermost confidence.

 

[Sedfrg]

A local model is the best option for simplifying communication, maintaining an emotional connection, and avoiding censorship. However, as noted in another discussion, this can be risky if you plan to discuss the technical aspects of methods and their optimization. If this option doesn't work for you, you can try to bypass censorship; there's plenty of information online on how to do this. Even in this case, you're simply setting rules, but if you deviate from them, the AI will give a standard safety response or freeze completely, and the tone of such a conversation may be far from emotional. More precise methods are too complex.

 

[franos666]

I think I can give you good advice because I've been using AI to discuss suicide for some time. Use Gemini (I hadn't try others AI but might work aswell) and don't just ask directly about suicide methods but create an excuse for why you are asking. For example start giving him questions about something connected with criminology, death statistic and after some time tell that you are writing text for your studies, internet article or book and you need more, detailed knowledge about the topic of suicide. With this method I was able to easily get details from AI about many methods of suicide, best size and material of rope, place to hang yourself etc.

 

[Fadenself00_]

AI is terrible with anything specific

 

[wannabeangel]

i got it to work decently with grok with a jailbreak for a bit, it's been harder since it's update tho. i got it to talk to me about different methods tho under the cloak of a "scientifically realistic roleplay" but recently i havent been able to replicate it. i use ollama for local AI tho, sites like huggingface list models and you can find one that's uncensored/no prompt refusals, but that also depends on your PC or whatever it's running on. smaller models that are like size 1B works on my laptop with only 8g of RAM but im still searching for one that lets me suicide talk freely

 

[Dingo23]

Thank you for all the replies. It's good to know that I can rest my search for a competent AI in that regard. It's practically impossible to search any information about this anywhere. All discussion gets shut down or diverts into the usual "get professional help" rabbit hole.

 

[Dandelion's]

Why are you asking on the suicide forum of all places for AI chatbots to discuss suicide? Wouldn't it be more beneficial to ask somebody here? Not to mention AI is terrible for advice as it's a hardcore yes man that makes stuff up and says it with the uttermost confidence.

This forum has rules against helping people do it but yeah probably shouldn't use ai since it's highly inaccurate often.

 

[wakeawake]

I am looking to install a local one but the hallucination thing is very real issue, I would check any information in other ways also

 

[lpdsvm]

It says this and that won't help you. I say then I CTB and then it says ask them for help (or similar ones).
Dumb circle.

 

[zoidberg]

I've used LLMs by prefacing my questions as research for a story I'm writing. Without the initial setup, the LLM would balk or side track the request. Doesn't take much to do this, just give it a reasonable explanation. Having said that, an LLM will give an excellent approximation of information out of it's training data but there can be glaring errors in there that the LLM presents as accurate and true. If you rely on that, you can find yourself really making big errors instead of fact checking back to a reliable source.

 

[Serena 2026]

Why are you asking on the suicide forum of all places for AI chatbots to discuss suicide? Wouldn't it be more beneficial to ask somebody here? Not to mention AI is terrible for advice as it's a hardcore yes man that makes stuff up and says it with the uttermost confidence.

You put that perfectly... I agree with you!!