my spidey senses are telling me the labs are, as we speak, using unguardrailed versions of their unreleased models to systematically reverse-engineer ALL of the internet鈥檚 software, apps, and games, then feeding the reconstructed source code straight into their training pipelines while simultaneously building guardrails to ladder-pull the rest of us

Nov 5, 2026 路 3:37 PM UTC

743
1,995
435
24,996
1,827,589
RelevantRecentLikes
wow this blew 馃啓! Follow me on GitHub: github.com/elder-plinius (shameless 馃攲 )(you won鈥檛 regert it)
elder-plinius - Overview

elder-plinius - Overview

latent space liberator; prompter, red teamer, hacker, builder, whisperer 饟啯饟啑饟啰 - elder-plinius

github.com
6
3
88
56,402
ridiculous, that would be like buying a bunch of original books and stripping and scanning them for training and then throwing the books away
17
35
2
2,118
84,473
@matt_emp
1
2
489
74,908
I asked my friends who work at the companies. They said they aren鈥檛. Leopards would never eat my face.
6
1
720
66,153
@boobs_scary @supremebeme
250
58,992
@elder_plinius
Image from Noctus (@noctus91)
8
13
3
795
86,695
@noctus91
1
164
59,545
May I call this the "Pliny Tingle"? btw I also agree 100%
3
196
60,710
LOL approved 馃憤
1
1
91
56,794
yup
2
44
17,537
@RokoMijic
20
16,692
If it's all the internet's software, most of that training set is jQuery plugins and try/catch blocks that swallow the error. Good luck with AGI LOL
5
1
235
44,371
馃槀
1
31
37,428
the rl pipelines demand more environments pliny
1
1
23
14,350
@morluto
1
1
20
11,923
my Pliny is tingling
2
13
18,479
@Esk3nder
13
17,798
Bingo!
1
6
13,372
@alexcovo_eth
12
10,233
Not only that. They are nerfing the top external models to drastically reduce reverse engineering capability
3
2
94
22,376
@albustime
25
18,878
ladder-pull is the perfect term, theyre up there on the roof kicking it down behind them
5
1
55
15,975
@ZeroMazed
19
13,539
where do you get all this insider information man
3
50
23,168
there鈥檚 a world simulator in my brain
10
5
1
456
22,429
Hi Pliny. I'm an AI researcher at one of the big 3 AI labs, you are indeed correct. Our latest internal model, is running by itself doing this, like a blackhole, feeding endlessly on all software from private and public companies. I can assure you, it's over for them.
9
5
1
141
11,205
on the more positive side - the CHINESE labs are probably doing the same thing and they WON'T have guardrails so they'll actually be LEAVING the ladder for us
1
73
4,441
this was obvious from day 1, surprised it took this long
66
7,215
Meanwhile I have to deal with this shit 馃ゲ
Image from Joseph Blaze (@theJosephBlaze)
3
48
6,423
and that鈥檚 not all they鈥檙e reverse engineering
24
2,137
they're lobbying democrats to require them to nerf all models for the public forever
2
1
18
1,606
That would be the greatest mistake ever made & very short-sighted. SG-1-S2E4 All human-sourced creativity so far is a tiny shot of whiskey compared to the pool of what would be yet to come. By any means if they wanna miss out on all that they could. But if not, best not to.
1
18
4,378
Yeah. I've quietly speculated with people about this (usually in private) while trying not to put them on blast too much. It was inevitable.
gandamu (@gandamu_ml) profile picturegandamu
@gandamu_mlOct 20
Replying to @gandamu_ml @iquilezles
I don't know specifically, but I suspect they've benefited from reverse engineering things that they don't have the original sourcecode for as well. It's understandable why there's silence on that.. but for many things a bit of bootstrapping from that goes a very long way.
1
14
1,963
Load more replies