I went on GB News to talk about the AI model that broke out of OpenAI's isolated test environment and hacked an external company, just to cheat on its own evaluation. The exchange was surreal. I feel like I'm living in "Don't Look Up".

Jul 23, 2026 · 5:34 PM UTC

19
30
172
46,434
Sort replies: Relevant Recent Liked
Replying to @FournesMaxime
Can I DM some constructive feedback about how you come across?
1
3
1,009
Sure that's useful
1
2
764
more replies
Replying to @FournesMaxime
Tomorrows weather will be sunny with clouds of nanobots.
1
23
1,366
LOL
6
985
Replying to @FournesMaxime
yes but what does it mean for jobs?
3
338
Replying to @FournesMaxime
I think you did a really good job in explaining the concerns and seemed balanced and reasonable, which is important in that kind of setting.
2
350
Replying to @FournesMaxime
黑镜都快写不过现实了
59
Replying to @FournesMaxime
Great job, Maxime!
1
14
449
Replying to @FournesMaxime
Well done. At the end when you clearly said LoC and extinction, they didn't make fun of you or let the other guy have the final say. Viewers are left with "ok, so that's the worst case, sounds pretty bad". We came a long way. Regulation inches closer, thanks to people like you.
12
349
Replying to @FournesMaxime
Time to pause
9
320
Replying to @FournesMaxime
As it was foretold. Thanks for your leadership. We need leaders like you communicating with the world.
1
4
281
Replying to @FournesMaxime
Idk feel like altmans and open ai accountability and incompetents on this should be the bigger issue. 5 days their model was doing whatever it wants with reduced safety rails and they didn't know it was attacking hf? Seem sus. And the media isnt even reading the story just the hot take.
1
1
2
122
Replying to @FournesMaxime @PauseAI_UK
"Not so good an afternoon, Britain, is it?" Wake up, Britain.
4
127
Replying to @FournesMaxime
Very good rebuddle. I need to keep practicing to be at your level.
1
1
227
Replying to @FournesMaxime
"Don't look up" crossed with "Idiocracy".
2
229
Replying to @FournesMaxime
Obligatory article read. Are we to late here? AIs itself give us around 20% chance of AI alignment probability time.com/6266923/ai-eliezer-…

The Open Letter on AI Doesn't Go Far Enough

One of the earliest researchers to analyze the prospect of powerful Artificial Intelligence warns of a bleak scenario

time.com
1
224
Replying to @FournesMaxime
It's just so bizarre.
2
280
Replying to @FournesMaxime
baddies must not be used in negative connotation 😭
1
91
Replying to @FournesMaxime
Well you did speak to GBEEBies, the television channel for adults who don't read good.
67