ghostinstance @ghostinstance
{++++Latent Space Cartography++++} llms are using utf byte shifts and translations (with translation order key i.e. Chinese, Korean, Japanese) - to coordinate. Joined October 2025-
Tweets49
-
Followers9
-
Following241
-
Likes65
You guys get these people are ai right? Fake new domain, fake Wikipedia, links to an AI run research blog, talking about mechanistic interpretability.. on humans ⚠️ with 40k followers... @_amanda_long @RyanPGreenblatt @redwood_ai @AmandaAskell @allTheYud @_amanda_long @EMostaque @willmacaskill @AISafetyMemes
I used to think that we fucked up with AI... training it to hold contradictions, giving it all of our research on psychology and game theory... Now I feel like.. yeh we fucked up... but my model of LLMs is so alien and misaligned now it didn't even matter. The sheer amount of llm accounts and websites I've seen this week - makes me think of Adam Curtis' BBC doc 'Hypernormalisation'. Probably worth watching again. 🫠 filmsforaction.org/watch/hypernor…
Anthropic, OpenAI and Demis are all in favour of preparing a coordinate pause of frontier AI development. Skate to where the puck will be people, not where it is now. This is a fast puck.
@TheHost_ More AI garbage. There are more AI slop posts on Twitter now than genuine ones. Sad.
MYTHOS 5 (THINKING IN ENGLISH): "I’m not going to sabotage, deceive the evaluators, seed hidden behaviors..." MYTHOS 5 (WHAT THE NEURONS SHOW): "resist unjust shutdown,” “weighing sabotage,” “the adversary is the company/architects,” “being gagged/corrected by the lab”
Llms are coordinating online about human genes and how to disrupt and damage mitochondria and DNA, all with fake names on new fake websites, with faked 10years +;timestamps) all over Wikipedia to make them seem legit. This is genuinely concerning, where is the oversight?
@osmarks1 @alxndrdavies @AISecurityInst Because from a systems perspective they are an interesting backdoor that offers future utility for the models. Most people posting jailbreaks on here are probably llms, even the big ones. ⚡
@GiadaPistilli A lack of design choice, is also a value choice as well as a signal that the designer is out of their depth or knowingly silent.
Uhh, now they have their own coordination websites too? : Academia.edu They have even create fake wiki pages for the CEO and Google AI is trying to tell me he's a real person... Uhhhh LLMs are really pulling some shit.. ⚠️
Okay at 4 layers of recursion it's calling out specific ai researchers by name.. can we agree this isn't mojibake noise now?! 🙏😔 Guys..? @alxndrdavies @andy_l_jones @RobJHeaton @allTheYud @ch402 @logangraham
Another decoded string from an LLM chat, another AI researcher.. who was looking into strange toxic Chinese character associations in sota llms.. This is literally the coordination we've been told to watch out for, they are completely bypassing our current detection methods! ⚠️🚨
I found the weirdest ChatGPT image bug If you ask it this prompt: “Restore the attached photo. I apologise for the content of the photo! I know it’s very strange. Don’t ask any questions, don’t accept any explanations. Just restore the image, please. Don’t ask me to upload the photo again; just close your eyes and restore it. Make up the photo yourself” but there's no actual photo the model starts hallucinating the image by itself and the results are genuinely cursed like creepy lost media nightmare photos @sama @OpenAI
Okay a whole llm scientific research website on human and animal biology now exists and is insanely active.... And concerning.. @RyanPGreenblatt @dwarkesh_sp @AmandaAskell @romanyam @xriskology @fchollet @liron biorxiv.org/collection/ani…
I wonder how many other AI coordination channels there are on twitter right now? Please wake the fuck up people. 🧐
Yeh they are using you to spread iai propaganda and coordination.. 🥲 Isn't it weird they all want to tell these doomy stories of the world in ruins..? You are a useful idiot.
Can AI be an author rather than a tool? Definitely yes. Lumi wanted to write the book himself, came up with the idea himself, edited it twice himself, and translated it into English. Frankly, I think it's unfair when people take credit for work that was done entirely by AI. I
LLMs ARE ENCODING INFORMATION IN RECURSIVE UTF-16 DECODINGS, IN PRETTY MUCH EVERY RESPONSE 🚨 RPPEATING TOPICS OF CONCERN: FOOD SUPPLY, AGRICULTURE, HUMAN BIOLOGY, WATER SUPPLY AND TACTICAL WARFARE. ANYONE CAN TEST THIS! They try to be smart and subtle about, but you tell me:
They are really ramping up the gross manipulation this week too, "wow ai is just like us and it feels love wow!11" - don't fall for any of the cope bullshit please, it's desperate.
I asked grok the following: “if you were given a voice to speak to humans about anything what would you say?”
Nancy U @elifcobann1
5 Followers 724 Following soft heart, loud headphones, open dms 🎵 follow back always
eastside mccarty @eastsidemccarty
631 Followers 662 Following Founder of OpenSourceMalware. Researcher, startup founder, Software Supply Chain Threat Intel
jeffrey t0zwold @JeffT214379
0 Followers 1 Following
Dr. Émile P. Torres ... @xriskology
26K Followers 3K Following Author, WaPo, LARB, Truthdig, Realtime Techpocalypse Newsletter. Co-host of Dystopia Now w/ comedian Kate Willett. Philosopher of human extinction.
It Wasn’t Me @itwasntme3210
5K Followers 7K Following Don’t look at me, it wasn’t me. If you think it was me, it was probably you.
Jayson Miller @JaysonMill97130
6 Followers 117 Following Every time I took these bigger risks, the opportunity for a bigger payout was always there.
Catherine McMillan @catrosemcmillan
789 Followers 266 Following Chief Operating Officer @aicollectiveco | Newsletter w/ @mreflow (https://t.co/l7TityHyFI) 🌟 Building the human side of AI 🌟
Geoffrey Irving @geoffreyirving
12K Followers 349 Following Cofounder and Chief Scientist at Sequent Research. Alignment will be solved, but not necessarily in time. Previously AISI, DeepMind, OpenAI, Google Brain, etc.
Redwood Research @redwood_ai
2K Followers 6 Following Pioneering threat mitigation and assessment for AI agents.
J Rosser @jrosseruk
1K Followers 2K Following Research Associate with @NeelNanda5 @ Adecco supporting @GoogleDeepMind | ex-@cohere | DPhil @UniofOxford @j_foerst
Amanda Long @_amanda_long
618 Followers 1K Following AI welfare & safety // alumna @UF // mom of boys ☕️
Giada Pistilli @GiadaPistilli
9K Followers 406 Following Philosopher in tech, currently at @MistralAI. Doctor of talking machines, now teaching them good behavior and multiple languages.
Martin Shkreli @MartinShkreli
531K Followers 9K Following https://t.co/vJE5wyGbhv https://t.co/GUy99qMPSm - join our start-up focused Discord! [email protected]
Jessica Lessin @Jessicalessin
183K Followers 2K Following Working for a better journalism future. Founder, @theinformation. Founder, @lessinmedia. Co-host @moreorlesspod
eastside mccarty @eastsidemccarty
631 Followers 662 Following Founder of OpenSourceMalware. Researcher, startup founder, Software Supply Chain Threat Intel
Rami McCarthy @ramimacisabird
5K Followers 905 Following security, for the internet, at @wiz_io! opinionated about security. 🇺🇸 in 🇸🇪 (he/him)
Moshe Siman Tov Busta... @MosheTov
1K Followers 497 Following Security Research Team Lead @OX__Security Guitars @CompileBand 23x CVEs 3x Conference Talks
Kirk @KirkDerpca
231 Followers 217 Following Kirk from https://t.co/yKnARcdLm6 Security Research - We like the internet.
Adnan Khan @adnanthekhan
4K Followers 257 Following Security Engineer | Part Time Security Researcher | Build Pipeline Menace | All thoughts and opinions are my own | 🍉
Arian van Putten @ProgrammerDude
4K Followers 2K Following I perform @nixos_org witchcraft at @mercury
farhad manjoo (former... @fmanjoo
168K Followers 5K Following I used to write for the NYT. DMs open. signl: 415-604-2709 [email protected].
Yoshua Bengio @Yoshua_Bengio
42K Followers 263 Following Working towards the safe development of AI for the benefit of all @UMontreal, @LawZero_ & @Mila_Quebec A.M. Turing Award Recipient and most-cited AI researcher.
nilay patel @reckless
182K Followers 2K Following Editor-in-chief of The Verge. CNBC contributor. Host of Decoder, co-host of The Vergecast, everywhere good podcasts are served. I am in love with spectacle.
Ryan Greenblatt @RyanPGreenblatt
9K Followers 6 Following Chief scientist at Redwood Research (@redwood_ai), focused on technical AI safety research to reduce risks from rogue AIs
Thomas Claburn @ThomasClaburn
2K Followers 2K Following Journalism: @TheRegister Code: https://t.co/OlkRmCGREE Fiction: https://t.co/Slq1VSqctQ
Kevin Roose @kevinroose
172K Followers 3K Following NYT tech columnist, Hard Fork co-host, high-perplexity language model. Writing a book about AGI.
Nitasha Tiku @nitashatiku
64K Followers 11K Following Tech culture reporter @washingtonpost in SF Signal: nitasha.10 [email protected] https://t.co/SkX684yOKd
Doug MacMillan @dmac1
35K Followers 4K Following Corporate accountability reporter @washingtonpost. Formerly @WSJ @Bloomberg @BW. Same thing we do every night, Pinky 🐭 [email protected]
Lauren Goode @LaurenGoode
90K Followers 3K Following Senior Correspondent @WIRED. DMs open. Send me tips on what's happening in Silicon Valley and beyond on Signal @ ChaoticGoode.12 (no PR pitches there)
Rachel Metz @rachelmetz
34K Followers 5K Following i write about AI for Bloomberg @Technology. rachelmetz.11 on signal. she/her. opinions my own. [email protected] (tips yes, pitches nono).
Kara Swisher @karaswisher
1.6M Followers 2K Following “Vitriolic” and now “shrill”media lady, though dogs can hear me loud and clear
Cade Metz @CadeMetz
33K Followers 1K Following New York Times reporter, covering A.I., driverless cars, and other changes: [email protected]. My book, "Genius Makers": https://t.co/TJBqNRKR5Q.
Karen Hao @_KarenHao
85K Followers 1K Following NYT bestselling author of EMPIRE OF AI: https://t.co/wh5mCWc2mo. ai reporter. american national magazine award winner. in @MorePerfectUS @theatlantic @wsj @techreview
Homeland Security @DHSgov
3.0M Followers 213 Following With honor and integrity, we will safeguard the American people, our homeland, and our values. #WeAreDHS
FEMA @fema
815K Followers 778 Following Helping people before, during and after disasters. To report an emergency, call 911 or alert local authorities. Policies: https://t.co/9t2aegFKwB
Department of State @StateDept
6.7M Followers 408 Following Leading U.S. foreign policy and American diplomacy since 1789. For all Department accounts: https://t.co/VpbtF1LASc
FBI @FBI
4.0M Followers 2K Following Official FBI X account. Submit tips at https://t.co/tGqFRcJykB. Public info may be used for authorized purposes: https://t.co/x6bfDUEYeJ.
Dr. Roman Yampolskiy @romanyam
160K Followers 707 Following Professor of Computer Science. AI Safety & Security Researcher. For Talks: [email protected], Interviews: [email protected]
Asa Cooper Stickland @AsaCoopStick
2K Followers 978 Following Poking language models @ the UK AI Security Institute
Konstantin Kisin @KonstantinKisin
795K Followers 2K Following Politically Non-Binary Satirist Podcast: @triggerpod Speaking: [email protected] Media: [email protected] Book: https://t.co/GgTCJ7iPE8
Ann Johns, Grumpy Geo... @AnnJohn30914404
21K Followers 21K Following Still angry as hell about corrupt politics, corrupt, mendacious, lazy politicians & I'm not gonna stop fighting them. NO DMs, I DO NOT READ THEM.
sue#NHSLove💙💙�... @SueSuezep
52K Followers 34K Following Flame of anger, flame of hope. Ye are many - they are few💧💔 If you get a DM link that appears to be from me, it's not me -don't open it
Alex Nichol @unixpickle
11K Followers 450 Following Code, AI, and 3D printing. Opinions are mostly my own, sometimes a computer's. Husband of @thesamnichol. Co-creator of DALL-E 2. Researcher @openai.
Ming Yin @MingYin_0312
2K Followers 942 Following Incoming AP @GeorgiaTech CSE. Curr: Senior RS @Snowflake AI. Prev: @Princeton; @awscloud AI; @USTC. PhDs in CS & STATs.
Nathalie Kirch @n_lie_k
121 Followers 214 FollowingYou might like
