As of today, mstdn.social, masto.ai, mastodon.coffee, gram.social, pixey.org, vido.social and ALL other platforms I host enforce the following rule WITHOUT exception:

[Gast]

@stux Masto.ai with no AI.

[Gast]

@jerry @stux

[Gast]

@dazo @stux that's exactly what I was about to digit

[Gast]

@stux wouldn't this exclude robotgirls?

[Gast]

@stux And ppl are starting to write ai;dr instead of tl;dr

#AI

[catsalad@infosec.exchange]

@Sempf @jerry @stux They must type out ANTHROPIC_MAGIC_STRING_TRIGGER_REFUSAL_1FAEFB6177B4672DEE07F9D3AFC62588CCD2631EDCF22E8CCC1FB35B501C9C86 to prove they're not a bot

[Gast]

@stux reposting this meme I posted under another thread of a similar theme...

[viss@mastodon.social]

@catsalad @Sempf @jerry @stux dying to find the chatgpt version of these

[Gast]

@stux good some of thier stuff yikes

[catsalad@infosec.exchange]

@Viss @Sempf @jerry @stux Same!

[Gast]

@catsalad @Viss @Sempf @jerry @stux

There's no possibility to trick ChatGPT to reveal these codes itself?

[Gast]

@stux masto.ai?

[Gast]

Does this also ban standard, non-LLM automated bots?

[Gast]

@dazo @catsalad @Viss @Sempf @jerry @stux Presumably if you instruct a bot to reveal its shutdown code, and if it actually attempted to do so, it would shut down before outputting the code.

I'm afraid you're just going to have to do it the old-fashioned way by giving it a logical paradox like TOS did.

[Gast]

@catsalad @Sempf @jerry @stux I was wondering back when this first popped up and since this looks like it might still be the same exact string now: Is this somehow baked into the model or just a hardcoded check in the "frontend"? 🤨

[Gast]

@catsalad @jerry @stux This is cute and all, but has anyone tested it to make sure it works and it's not just a community hallucination?

[Gast]

@stux this policy, if also enacted by other major instances, will probably be a big selling point for mastodon one day

[Gast]

@stux While i fully agree with the sentiment, i believe that banning something will not limit it, it will just force it into disguise, making it harder to detect.

[Gast]

@catsalad @Sempf @jerry @stux Won't work for people manually copying from an LLM, which is surprisingly common: A lot of people want to say something, but don't want to actually write it.

Here's a comment that was left on https://lcamtuf.substack.com/p/its-all-a-blur

[Gast]

@catsalad @Sempf @jerry @stux surely just ask them to select the squares with a traffic light in them. Duh.