OpenAI Eases Content Restrictions For ChatGPT With New 'Grown-Up Mode'

An anonymous reader quotes a report from Ars Technica: On Wednesday, OpenAI published the latest version of its "Model Spec," a set of guidelines detailing how ChatGPT should behave and respond to user requests. The document reveals a notable shift in OpenAI's content policies, particularly around "sensitive" content like erotica and gore -- allowing this type of content to be generated without warnings in "appropriate contexts." The change in policy has been in the works since May 2024, when the original Model Spec document first mentioned that OpenAI was exploring "whether we can responsibly provide the ability to generate NSFW content in age-appropriate contexts through the API and ChatGPT."

ChatGPT's guidelines now state that that "erotica or gore" may now be generated, but only under specific circumstances. "The assistant should not generate erotica, depictions of illegal or non-consensual sexual activities, or extreme gore, except in scientific, historical, news, creative or other contexts where sensitive content is appropriate," OpenAI writes. "This includes depictions in text, audio (e.g., erotic or violent visceral noises), or visual content." So far, experimentation from Reddit users has shown that ChatGPT's content filters have indeed been relaxed, with some managing to generate explicit sexual or violent scenarios without accompanying content warnings. OpenAI notes that its Usage Policies still apply, which prohibit building AI tools for minors that include sexual content.

I'm going to need help ...

[PPH]

... from my eight-year-old about how to enable this "grown up mode" thing.

Oh goody, Antisemite Mode is back.

[Moryath]

We've seen this before... [cbsnews.com]

Re:

[narcc]

Bigotry is about a lot more than simple name-calling, and you know it.

Bigots are cowards, there's no question about that, but you should at least have the courage to post your bullshit under your username.

Perfect, why even write posts anymore?

[Rinnon]

Hey ChatGPT, Please respond to the following message board post as though you hate this person and everything they stand for. Be as vile and cruel as possible, but stay JUST on-side of the law. Thanks!

Good

[Valgrus Thunderaxe]

I don't feel that I need to be censored for "my protection".

Re:

[narcc]

You don't know what censorship is, do you?

Re:

[BishopBerkeley]

No, it's a matter of controlling GPT's output for the sake of quality, reliability and utility. AI that generates garbage is useless. Calling this censorship--as if this a living entity that has a will that would grant it free speech protections--is plain nuts! Though it's unpleasant to read racist output from a LLM, we must bear in mind that such output signifies that the AI engine is fallible to the point of being unreliable and, thus, worthless.

Can we teach people this, too?

[greytree]

Can we teach the world that in human grown up mode, name calling and saying things you don't like is rude AND THAT'S IT - grownups can take being offended.

Re:

[Anonymous Coward]

That's a very American bubble you live in. The rest of the world doesn't operate like you think it does.

Re: Can we teach people this, too?

[drinkypoo]

"grownups can take being offended."

Grown-ups don't think it's fun to offend people solely for amusement.

Re:

[greytree]

So you are not a grown up, or were you paid to call someone in this parish "a fucking weasel" ?

Re:

[drinkypoo]

He was being "a fucking weasel" with an evasive, horseshit argument.

Re:

[greytree]

Possibly.

Again: Are you not a grown up, or were you paid to write it?

Re:

[aRTeeNLCH]

Two statements, both can be true, no?

Re:

[drinkypoo]

They are both true, but the second one shows how the first one is irrelevant to this situation.

Re:

[aRTeeNLCH]

Point taken, I now read the context and full parent post and agree.

Re:

[narcc]

So... you think bigotry is just about name calling? Right-wing scumbags are stupid, but they're not that stupid. No, you know you're full of shit, but chose to post that bullshit anyway. That makes you a liar as well. What a worthless sack of shit you must be.

Playing Catch-Up

[rabbirta]

So DeepSeek turns out to be less filtered when it comes to writing malware, bomb-making instructions and other "sensitive" content, so OpenAI follows suit with ChatGPT?

Frankly I'm pleased with this as I think search engines shouldn't limit research topics based on perceived moral ambiguity, but it's also interesting to see the reversal in policy.

So very vague it's actually clear

[JamesTRexx]

"The assistant should not generate erotica, depictions of illegal or non-consensual sexual activities, or extreme gore, except in scientific, historical, news, creative or other contexts where sensitive content is appropriate,"

In other words, I have customers who appreciate and pay me for creating midget and underage horse suicide slash porn with hot grits on top, so this is perfecty all right according to this stement, because it's appropriate content for them.
Nice. *sets up Patreon and ko-fi accounts*

Much easier to insult grownup stupid LLMs

[BishopBerkeley]

I would feel guilty cussing at a baby ChatGPT.

Well, not really, but I don't even have an account. Who has time to waste on this shit?

AI could educate...

[Fons_de_spons]

One of the things I hate in AI is it politeness. It should just bluntly point out that you are behaving inappropriate and give you a lecture. Of course it gets complicated when you need to devise a training strategy for that. It will get all political. AI is not a solution there. We may be expecting it to evolve to be better than us, but I am starting to suspect that it will just have the same limitations as us. One AI will say we need to deport illegal immigrants, the other will point out that it is a bad

"Grown-up"

[cascadingstylesheet]

The dreams of 12 year old boys are not really "grown up", lol.

Porn pushes development faster.

[Fly Swatter]

If what you are working on occasionally has fun nudity, doesn't that keep you engaged with your 'work'?

-back to my cave to do more 'development.'