curiosity. technology. art. memes.

/mnt/c/
Joined July 2020
Never apologize unnecessarily. Never be apologetic. Always act with respect. Never appeal to authorities. Never plead. Always invalidate accusations. Never explain yourself preemptively. Never interrupt self-destruction.
1
6
407
Michael Vu retweeted
Lilas Ikuta, la famosa vocalista de YOASOBI conocida como ikura, dejó en shock a sus fanáticos tras armar un concierto IMPROVISADO en medio de un parque en Sapporo armada únicamente con su guitarra. Luego de insinuar en redes que el clima estaba bonito para cantar, la artista interpretó hits como Yoru ni Kakeru ante la mirada de los transeúntes que corrieron al lugar para verla en vivo. Se trató de su primera presentación callejera en OCHO años, reviviendo la etapa de cuando tocaba en la vía pública a los 14 años antes de volverse una estrella global y justo antes de arrancar su nueva gira masiva de domos.
47
2,286
94
20,982
536,229
Hi, I'm Felix. I'm an eng lead working on Claude. I love watching people take flight and do things they never thought possible. We often describe that as "being more ambitious" when using AI, which is tricky advice to follow. I wrote a tutorial: felixrieseberg.com/how-to-be…
395
2,472
550
23,793
5,229,311
“The romantic fantasy for a lot of men isn't a relationship constantly being tested from within. It's a relationship strong enough to withstand everything testing it from outside.”
15
417
4
3,875
23,154
One paragraph from C. S. Lewis can change your life
119
3,004
108
22,177
684,340
Michael Vu retweeted
OpenAI’s post is very very very carefully worded so here’s more information from the criminal complaint and DOJ statements. The individual that planted the device was Abdulrahman Alfozan, 35, a Saudi national and engineer on the product team who joined OpenAI in July 2025. Key allegations from the complaint (N.D. Cal.): • On or about Sept. 30, OpenAI’s technical surveillance team found a surreptitious audio recording device plugged into a USB port on the back of a TV in an executive conference room. It was concealed in toilet paper, installed around Sept. 28, and its contents were encrypted. • The room is used for highly confidential technical, commercial, and strategic discussions. Alfozan had no authority to access it or attend meetings there. CCTV showed him near the room on multiple dates beforehand. • On Oct. 5 he entered the room and appeared to look behind the TV. That same day he left an unattended bag in another room containing his wallet, ID, and a second similar recording device. • Before installation, he queried internal tools about building addresses, meeting rooms, room capacities, calendars, and locations of leadership and external partner meetings. • A chatbot account tied to his spouse (sharing multiple identifiers with his own) submitted prompts between Sept. 24–30 about building a continuous audio recorder, encrypting the SD card, concealing the device, adding Wi-Fi/remote access, and the specific device’s MAC address. He’s been charged with using or attempting to use a device to intercept oral communications (18 U.S.C. § 2511). If convicted: up to 5 years, $250,000 fine, and 3 years supervised release. Arrested Oct. 9, initial appearance Oct. 13, remains in custody. Detention hearing set for Oct. 19. OpenAI says it used its models to help identify him, coordinated the termination with the FBI (access was progressively restricted beforehand), and has no indications of model-weight, source-code, or customer-data exfiltration. U.S. Attorney described it as a national-security matter; FBI has flagged protection of American AI innovation as a high counterintelligence priority. Friendly reminder that physical + digital insider-threat controls matter even when the goal isn’t classic IP exfiltration. Wiretap charges don’t require successful data theft. Toilet-paper concealment and a second device left in a bag with ID are not sophisticated tradecraft. (Complaint details drawn from SF Standard reporting, DOJ/USAO Northern District of California statements, and OpenAI’s own post.)
We discovered a sophisticated recording device that was hidden inside one of our offices. We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier. We referred the matter to the FBI, and we’re grateful to the agents and prosecutors who moved quickly to make an arrest. The FBI’s investigation is ongoing and we will continue to support their work. At this time, we have no indications of this individual inappropriately accessing or exfiltrating model weights, source code, or customer data. The employee was terminated. We take protecting our technology, data, and people seriously, and as a matter of national importance. Our security program is designed to protect against a wide range of threat actors, from insider risks to cybercriminals to nation states, and this case is an example of how AI can help address these sophisticated security threats.
88
182
53
2,436
629,464
YouTubeショート見てたら流れてきたこれが面白すぎて今日はもう何も出来ない
77
2,876
87
31,814
993,227
Michael Vu retweeted
you fell in love with what you thought we could be. that’s not really my fault
3,189
613
2,627
13,666
13,736,829
Michael Vu retweeted
34
129
13
2,081
79,571
Michael Vu retweeted
Marriages survived because men had the heart to endure women. It lasted for generations because of MEN. Because of our emotional intelligence and our compassion for our wives. Not because women were good partners. Because men learned to overlook a lot of stupid shit women do just to let peace reign. Let's be real, women are NOT easy to live with. They are not romantic. Their whole idea of romance is RECEIVING. What he bought for me, where he took me, what he did for me. Never what they give. They are not funny. They are overly sensitive, they hate accountability, they hate saying sorry, they hate initiating sex with their own husbands, and they are controlling as hell. Stubborn and rebellious. And that domestic violence you hear about in marriages? Most of it is STARTED by the wife. Pushing, slapping, throwing things, talking reckless, provoking until the man reacts. Then she wants to play victim. Look at any relationship that doesn't have a man in it,it doesn't work. Lesbians? Highest divorce rate. Highest abuse rate. Because there is no man there to tolerate their violence and their emotions. Two women together can't handle each other. Female friendships? Even worse. Pure gossip, backbiting, jealousy, cutting each other off over the most childish things. They will have 10 female friends by 2022. By 2026, only 2 are left. She has fought with ALL of them. Every single one is now an enemy. But men are the problem.
what opinion on relationships will have you like this?
2,008
12,220
969
78,716
3,596,182
Michael Vu retweeted
This seems very not good
Two other safety researchers and I were fired from OpenAI last week. We wrote this letter to leadership. I believe we were fired for prioritizing safety over the near-term interests of OpenAI as a corporation.
49
23
1
1,080
92,788
Michael Vu retweeted
this is so embarrassing i’m in tears😭😭
#France — If you don’t run when the police charge at you, they will retreat.
447
13,584
99
219,156
4,393,018
Michael Vu retweeted
We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks: Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better: Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better: Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better: Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work! In summary: - As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding. - Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.
1,652
6,603
1,525
55,770
7,928,598
Michael Vu retweeted
İşte bu nedenle filtrasyon, bir akvaryumun en önemli parçalarından biridir.
17
102
7
2,608
1,094,280
Doing business in Australia 🤦🏽‍♂️ Business Owner: You can’t do that! ATO: Watch me!
102
579
82
2,524
117,385
W excuse to share this
Women when men lock their dicks
49
1,141
56
16,083
661,329
魔法少女シリーズ #ReSURGUM
112
10,947
121
100,222
1,726,060
Michael Vu retweeted
im actually obsessed
44
1,250
86
27,501
3,299,247
Key Animation: Yn
2
236
4
3,440
63,191
You need to have the knowledge of the gods to draw a hand and rotate it in 3D space like that
im actually obsessed
33
997
10
33,256
1,296,162
Michael Vu retweeted
Vivimos en una sociedad en la cual una persona la cual se queja cada día de que no tiene privacidad patrocina unas gafas que invaden tu privacidad
LISA stuns for new Meta glasses collaboration.
106
6,957
102
88,065
1,226,394