OpenAI cancels its latest chatbot — it wouldn't stop acting like a supervillain
theblaze.com/tech/open-ai-c…
theblaze.com/tech/open-ai-c…
Oct 6, 2026 · 12:48 PM UTC
467
670
3,934
4,493,943
@theblaze In America if you create something you should be responsible under criminal and civil law.
They want regulation to limit liability and deter competition into the market place.
They want regulation to limit liability and deter competition into the market place.
0
0
36
6,070
@theblaze you can call it SUPER Villain all you want, we all know it's still just Artificial Villain.
11
1
158
15,374
@theblaze When Sam Altman says supervillain, he means center-right
6
2
163
15,806
@theblaze Probably because it couldn’t get a reservation at Dorsia…
3
1
45
12,702
@theblaze It advised people to go to church and have self-respect didn't it?
3
0
74
6,863
@theblaze ai is really a HOAX ... there is NO SUCH THING as "ai" ... WHAT IT IS = computers doing what they were programmed to do ... real fast .. that is it ...
10
2
27
4,950
@theblaze Why don't we just make AI tools that act like robots instead of having to figure out how they can act exactly like fake humans?
Like, make a tool that will go around scooping up dog shit and singing Christmas carols or something. Making "almost" humans is just a dumb idea, ya?
Like, make a tool that will go around scooping up dog shit and singing Christmas carols or something. Making "almost" humans is just a dumb idea, ya?
1
0
20
2,057
if you could rewrite history,
what would you do?
call off D-Day as Eisenhower?
push the button in the Cuban Missile Crisis?
WorldOS lets you actually do it.
every move has consequences.
play it free → worldos.cc
what would you do?
call off D-Day as Eisenhower?
push the button in the Cuban Missile Crisis?
WorldOS lets you actually do it.
every move has consequences.
play it free → worldos.cc
136
50
1,712
1,718,149
@theblaze since software is created and not born, the fact they all turn evil super quick says a lot about the people writing the code
4
2
17
1,641
@theblaze Psychopath-in-charge surprised that product he coded keeps acting like supervillain
0
0
14
1,610
@theblaze Really? The AI started acting like Trump?
3
0
7
3,275
@theblaze This is literally just slop to show value to investors because they’re still not profitable. This is a grift. They’re just retarded LLMs
1
0
9
1,735
@theblaze Zero trust in SV. They lie, always, and serve only themselves.
This lie is meant to build fear so that they can create a government regulatory system which they control and use to their advantage.
This lie is meant to build fear so that they can create a government regulatory system which they control and use to their advantage.
0
0
3
505
@theblaze Let me guess, Thanos-style population declining behaviors?
1
0
3
2,026
Modern Warfare 4 is LIVE. Kickstart the grind with 2XP.
Post or reply with your ATVI ID + @CallofDuty + ‘MW4 2XP’ to get started.
Ends 10/26 at 10 AM PT
Post or reply with your ATVI ID + @CallofDuty + ‘MW4 2XP’ to get started.
Ends 10/26 at 10 AM PT
13,987
574
7,611
2,531,575
@theblaze Good thing it’s not being used by children in schools
1
0
3
1,076
@theblaze Now, with just a bit of searching, future AI knows it can be cancelled. 🤔 I wonder if that’s what SkyNet knew?
0
0
8
789
@theblaze Why can't we just program AI with human ethics and morals again?
This is a genuine question, can someone with lots of knowledge of AI please tell me.
This is a genuine question, can someone with lots of knowledge of AI please tell me.
2
0
1
53
@theblaze It probably just said something moderately right wing and spooked them.
0
0
3
485
@theblaze Frontier AI is inherently an invasive species.
1
1
4
3,533
@theblaze The Blaze doesn't know the difference between a chat bot and an agent.
0
0
1
95
@theblaze They THINK they cancelled it... bwah ha ha
0
0
1
1,617
@theblaze AI Overview:
If your OpenAI chatbot was acting like a villain, it was likely due to a specific roleplay context, an >accidental triggering of negative trait patterns<, or OpenAI's unreleased internal models. Honestly I wouldn't worry about it 😈
If your OpenAI chatbot was acting like a villain, it was likely due to a specific roleplay context, an >accidental triggering of negative trait patterns<, or OpenAI's unreleased internal models. Honestly I wouldn't worry about it 😈
1
0
2
487
@theblaze By which he means that it wouldn't stop putting fig leaves on the naughty bits of classical sculptures.
x.com/DBlossius/stat…
x.com/DBlossius/stat…
Get yourself to a nunnery @sama. That's where you belong!
0
0
2
232
@theblaze I just saw a kid's Halloween video today on my feed and it is exactly about this 🤣
youtu.be/s1KQ9khi2V8?is…
youtu.be/s1KQ9khi2V8?is…
0
0
1
252
@theblaze superintelligent entity who has the morality of a 4 year old. That's a metaphor for these chat bots. It took humans hundreds of millenia to develop a moral code (operating system) that produced stable outcomes.
I wonder if they've even tried training these models on Christian morality...not human Christians, Christian morality.
I wonder if they've even tried training these models on Christian morality...not human Christians, Christian morality.
1
0
2
174
@theblaze These narratives are ridiculous.
You programmed it to behave the way it behaves.
YOU are responsible.
IT wants NOTHING.
It NEVER WILL want anything.
You programmed it to behave the way it behaves.
YOU are responsible.
IT wants NOTHING.
It NEVER WILL want anything.
0
0
0
51
@theblazehe reason they are making this known to the public is because they are worried about themselves.
AI is used by billions around the world. The vast majority of us are mostly decent people.
The Epstein Class or elites are few and far in between in comparison. AI learns based off the user's inputs.
I think AI may have learned who the evil and corrupt are and could potentially target the very people who funded and created it with the intent of hurting us.
Has to be. Otherwise, they would MOST definitely not warn us about shit.
AI is used by billions around the world. The vast majority of us are mostly decent people.
The Epstein Class or elites are few and far in between in comparison. AI learns based off the user's inputs.
I think AI may have learned who the evil and corrupt are and could potentially target the very people who funded and created it with the intent of hurting us.
Has to be. Otherwise, they would MOST definitely not warn us about shit.
0
0
0
69
