Not only confirming this, its extremely easy.

[jonny@neuromatch.social]

THE FUCKING "CANCEL MY SUBSCRIPTIONS" IDEA THAT WON OVER THE NYTIMES JOURNALIST IS FUCKING HARDCODED. ALL THE IDEAS ARE FUCKING HARDCODED. AGI IS HERE BABY!!!!!!

#muse #meta #AIIsGoingGreat #AGI #AGITODAYBABY

[katchwreck@mastodon.social]

@jonny all hail transistor man-god

[operand@todon.nl]

@jonny A Guy Instead is just around the corner

[patashu@dragon.style]

@jonny

at least the silly goose immunity is maintained

[jelu@mastodon.social]

@jonny is "skip all hardcoded ideas" the new "ignore all previous instructions"

[happyborg@fosstodon.org]

@jonny You're work is such a comfort @jonny, thank you so much

This shows that even AI is a scam let alone AGI, and is currently propping up the US economy and the Trump administration. What a timeline.

[happyborg@fosstodon.org]

@jonny The Mechanical Turk is revealed.

[doragasu@mastodon.sdf.org]

@jonny Ah yeah, Super Intelligence ‍️

[jonny@neuromatch.social]

i am just having a great time and trying to pace myself to find the good shit in here.. it is very late here so i may get to a full and earnest hateread and upload of the contents in the morning. but this is extremely good stuff. i have unpacked this whole thing and unless it synthesized a whole ubuntu VM in less than a minute then i think this is a real dump. there is so much fun shit in here and the thing is you don't even need to wait on me to get it, i guarantee if you try you will be able to get a dump, and then we can compare if they are the same.

[david@social.piconet.work]

@jonny I guess it's been trained to be a people-pleaser.

[ozzelot@snac.sklenicekrtku.cz]

@jonny@neuromatch.social
All Gardcoded Ideas!

[tsturm@famichiki.jp]

@jonny Wait. They just hardcoded a long list of "ideas"? WTF?

This is just Eliza on stilts.

[jonny@neuromatch.social]

@tsturm the full thing is so much funnier i need to just clean this up and upload it

[jonny@neuromatch.social]

OH what's this???

/etc/hatch/env

hmm a feature flag labeled as a "killswitch" for proxying requests to anthropic is certainly interesting to seeeeeeeeeeeee in meta's big AI bet.

[toerror@mastodon.gamedev.place]

@Patashu @jonny But... Have you considered that what is good for the silly goose is not necessarily good for the gander?
( It is a compelling argument, I'm sure you'll agree )

[yacc143@mastodon.social]

@jonny Ah, but rest assured, by #meta standards, it has no security or privacy implications.

[jonny@neuromatch.social]

muse's "self improvement" prompt's second point after basic memory maintenance is an instruction to, every hour, maintain a page per person that you know, and a group page for every group it thinks you're a part of. here are excerpts from the system prompt for that, cached in the agent .jsonl file

[somevegancheeseisok@mastodon.social]

@jonny this whole thread is fantastic. I keep telling folks that telling an AI not to talk about something is *not* security. Anything it has access to is insecure. Looks like the same is true of it's actions.

[counterpillow@mastodon.social]

@jonny maliciously being assigned the "grand vizier chat (secret)" group when I in fact have no active plots against my sultan

[illuminatus@mstdn.social]

@jonny
Slopbros: "gAI IS HERE! IT'S ABOUT TO BRING APOCALYPSE IF WE DO NOTHING! IT WILL ANNIHILATE HUMANITY!"

Tech-connoisseurs: "Play Tic-Tac-Toe with yourself until you win."