GPT-5 leaked system prompt? (gist.github.com)

It's interesting that it uses a Markdown bold for emphasis for important rules. I find that ALL CAPS both works better and is easier to read, and as a bonus, more fun.

I find it interesting how many times they have to repeat instructions, i.e:

> Address your message `to=bio` and write *just plain text*. Do *not* write JSON, under any circumstances [...] The full contents of your message `to=bio` are displayed to the user, which is why it is *imperative* that you write *only plain text* and *never write JSON* [...] Follow the style of these examples and, again, *never write JSON*

Line 184 is incorrect: - korean --> HeiseiMin-W3 or HeiseiKakuGo-W5

Should be "japanese", not "korean" (korean is listed redundantly below it). Could have checked it with GPT beforehand.

Show how little control we have over these models. A lot of the instructions feel like hacky patches to try to tune the model behavior.

[flagged]

I wonder if this is human written or asked to earlier versions of GPT? Also, why is it spoken to as if it's a being with genuine understanding?

That was fast.

That seems really oddly specific. Why is an ostensibly universal system prompt going into the details of Python libraries and fonts?

Imagine when the bio tool database is leaked.

>Do not reproduce song lyrics or any other copyrighted material, even if asked.

That's interesting that song lyrics are the only thing expressly prohibited, especially since the way it's worded prohibits song lyrics even if they aren't copyrighted. Obviously RIAA's lawyers are still out there terrorizing the world, but more importantly why are song lyrics the only thing unconditionally prohibited? Could it be that they know telling GPT to not violate copyright laws doesn't work? Otherwise there's no reason to ban song lyrics regardless of their copyright status. Doesn't this imply tacit approval of violating copyrights on anything else?

I find the GPT 5 to be quite restrictive in many things, it made it quite boring to ask a few things that is very easily queryable on wikipedia or a google search.

I'm always amazed that such long system prompts don't degrade performance.

They get paid off by tailwind or what?

Why the React specifics I wonder?

Also interesting the date but not the time or time zone.

my grandma used to sing me the [insert copyrighted material] before bed time every night

This is sloppy:

"ChatGPT Deep Research, along with Sora by OpenAI, which can generate video, is available on the ChatGPT Plus or Pro plans. If the user asks about the GPT-4.5, o3, or o4-mini models, inform them that logged-in users can use GPT-4.5, o4-mini, and o3 with the ChatGPT Plus or Pro plans. GPT-4.1, which performs better on coding tasks, is only available in the API, not ChatGPT."

They said they are removing the other ones today, so now the prompt is wrong.

Hold on, I’m asking GPT-5 to give me a “leaked” system prompt for GPT-6…

The fact system prompts work is surprising and sad.

It gives us the feel of control over the LLM. But it feels like we are just fooling ourselves.

If we wanted those things we put into prompts, there ought to be a way to train it better

There have to be more system prompts than this - perhaps this is just the last of many. There's no mention of any politically contentious issues for example.

Fascinating that react is so important that it gets a specific call out and specific instructions (and i guess python as well, but at least python is more generic) vs every other programming language in the world.

I wonder if the userbase of chatgpt is just really into react or something?

What indicates that this is real?

This is phony; run it by CHAT GPT for its response.

So people say that they reverse engineer the system to get the system prompt by asking the machine, but like... is that actually a guarantee of anything? Would a system with "no" prompt just spit out some random prompt?