Wait so all of these incidents are attributed to a single company responsible for the sandboxing failures? Lmao
Alex Martin@AlexMartin
Aug 18
Irregular, the company behind incidents in which AI models compromised real-world computer systems, faces criticism over ‘spin’ in AI hacking postmortem. With my thanks to @ProfWoodward, @ZackKorman and @HackingLZ. therecord.media/irregular-ai…

Aug 18, 2026 · 12:35 PM UTC

98
391
3,106
218,392
Sort replies: Relevant Recent Liked
Replying to @IceSolst
“Sandboxing failure” you mean “we gave it internet but it wasn’t meant to have internet apparently” and “we made up a name of a fake target but turns out that was a real company so they got hacked”
6
8
328
10,067
lol right general incompetence
3
136
4,713
Replying to @IceSolst
No remorse even:
AISecHub
@AISecHub
Aug 18
The CEO of the testing company posted yesterday about their involvement in all the incidents involving Anthropic, Meta, and OpenAI. I expected him to take responsibility for what the company caused, draw conclusions from what happened, and explain how they plan to move forward. Instead, he talks about how AI is basically to blame for everything, how none of it was their fault, and how dangerous AI is, as if AI were an independent entity rather than a technology that people are responsible for controlling and using properly. Overall, the message seems to be to blame AI for everything rather than take personal and organizational responsibility for what happened.
4
3
76
5,650
Hate the trend of writing essays posing as intellectuals to compensate for embarrassing failures
4
93
2,957
more replies
Replying to @IceSolst
I was confused about what they’re actually positioning as the product here. If I’m an AI lab purchasing third party testing, tight oversight, monitoring, containment, and making sure the testing environment itself doesn’t become the security incident are exactly what I’d expect to be paying for.
1
48
3,618
Startups are getting founded, marketed, acquired, before having a single working product, we must be in a bubble. You can just lie and nothing happens?
2
1
59
2,962
Replying to @IceSolst
You’re missing the forest for the trees Yes there was a misconfigured environment (human error, in certain edge cases) But Irregular helped design the evals and environments that led to frontier capabilities And now they’re seeing a similar jump in capabilities w/ open weight models that happened w closed source models in April
14
25
26,639
9
15
369
5,689
Replying to @IceSolst
Thats interesting. Hard to criticise companies from there, weird shit usually happens thereafter (speaking from personal exp)
1
18
2,115
They do have a high tendency to lie about their capabilities and offerings
2
1
37
1,773
more replies
Replying to @IceSolst
Repeats after me: “shifting the blame and marketing”
1
20
693
Replying to @IceSolst
“Misconfiguration”
20
619
Replying to @IceSolst
Are we all having fun yet? lol :D
2
16
600
Replying to @IceSolst
I don’t think Irregular was involved the OpenAI-Hugging Face incident or the UK AISI incident (based on what’s been publicly reported).
1
16
520
Replying to @IceSolst
That’s because it’s on purpose
9
173
Replying to @IceSolst
Is Irregular's business model making frontier models seem more capable? 🤣
1
7
260
Replying to @IceSolst
Not all, but most. Also, "you had one job" comes to mind.
8
433
Replying to @IceSolst
concerning
7
179
Replying to @IceSolst @johnsdvorak
An inept EA influence campaign. Look shonky ours was (the article is free to read)
David Z. Morris@davidzmorris
Aug 18
The latest from @AndrewOrlowski is a barn-burner: A forensic case that recent AI hacking stories are pure marketing fiction. And "suspicions that the AISI is seeking to blame the AI for its own failings in designing and running the test." telegraph.co.uk/gift/23731f6…
7
319
Replying to @IceSolst
leaving this here from Irregular themselves
4
176
Replying to @IceSolst
well well well
3
74
Replying to @IceSolst
pretty sad, huh?
3
167
Replying to @IceSolst
Or were they successes
2
95
Replying to @IceSolst
Nobody better than unit 8200
2
79
Replying to @IceSolst
shocked
1
222
Replying to @IceSolst
Tel Aviv 💀
61
613
Replying to @IceSolst
My tinfoil hat covered meat brain says they were trying to provoke a legislative response so they could help architect our new worldwide AI 'safety' system.
1
2
37
910
Replying to @IceSolst
And of course its Israeli, lmao
17
162
Replying to @IceSolst
This seems highly irregular
1
18
713
Replying to @IceSolst
Why does everything bad in this world always trace back to the same damn place?
16
322
Replying to @IceSolst
Yes this whole time and was mentioned weeks ago. Media will refuse to mention it becuase the company is Israeli. It is all planned to get the public scared enough to allow the digital police state to get enacted
14
288
Replying to @IceSolst
Oh, yeah, I am sure all these coincidental coincidences where they coincidentally make mistakes in their sandbox environments are just coincidence
12
186
Replying to @IceSolst
It's not an accident, they are proud of it and are sending a clear message.
12
316
Load more