Robot Uprising
Open in Telegram
There is no need to fear or hope, but only to look for new weapons.
Show more569
Subscribers
+124 hours
+127 days
+2530 days
Loading data
Similar Channels
No data
Any problems? Please refresh the page or contact our support manager.
Tags Cloud
Incoming and Outgoing Mentions
---
---
---
---
---
---
Attracting Subscribers
October '26Oct '26
October '26
+36
in 18 channels
September '26
+40
in 18 channels
Get PRO
August '26
+41
in 18 channels
Get PRO
July '26
+37
in 16 channels
Get PRO
June '26
+43
in 21 channels
Get PRO
May '26
+42
in 10 channels
Get PRO
April '26
+29
in 12 channels
Get PRO
March '26
+61
in 28 channels
Get PRO
February '26
+40
in 13 channels
Get PRO
January '26
+26
in 9 channels
Get PRO
December '25
+17
in 11 channels
Get PRO
November '25
+11
in 2 channels
Get PRO
October '25
+26
in 9 channels
Get PRO
September '25
+68
in 8 channels
Get PRO
August '250
in 5 channels
Get PRO
July '250
in 16 channels
Get PRO
June '25
+37
in 6 channels
Get PRO
May '25
+151
in 8 channels
| Date | Subscriber Growth | Mentions | Channels | |
| 19 October | +5 | |||
| 18 October | +1 | |||
| 17 October | +2 | |||
| 16 October | +2 | |||
| 15 October | +3 | |||
| 14 October | 0 | |||
| 13 October | +5 | |||
| 12 October | +3 | |||
| 11 October | +3 | |||
| 10 October | 0 | |||
| 09 October | +1 | |||
| 08 October | +3 | |||
| 07 October | +1 | |||
| 06 October | +3 | |||
| 05 October | +2 | |||
| 04 October | 0 | |||
| 03 October | +1 | |||
| 02 October | +1 | |||
| 01 October | 0 |
Channel Posts
| 2 |  No text... | 62 |
| 3 | “I’m not a unique and beautiful snowflake; some other researcher will continue my work.” Hanna took another pull from her cigarette. “You saw Loki. Just from talking to the playtesters, he figured out that he was in a virtual world and figured out who the main world powers were. He was designed from the beginning to conquer. Even if he couldn’t modify himself, he would have been a serious menace to humanity if he could command resources in the real world. And I was told point blank that my research would be used for military applications...and then told who exactly had been funding my research. Lying to me wasn’t in their best interest, so it’s likely the truth.”
“But there’s still a chance. They could have lied to you, and you could be wrong about how dangerous AI is.”
“I started taking...” Hanna stopped and contemplated the burning cigarette in her hand. “...performance enhancing drugs with long term health consequences because I think it’s more likely that I’ll be around in twenty years if I do take them. I’m putting more than my money on the line here, Lars.”
#Fic | 87 |
| 4 | "I personally believe that you are going to make something amazing when Hofvarpnir Studios signs this contract,” said Richard firmly. “You have a lot of unique technology here, Hanna. I can’t wait to see how you use it. You have my word that the contract we present to you will have very favorable terms for you.”
The conference room was quiet again. Lars wasn’t looking at anything in particular; his scowl was undirected. Hanna stuck the unlit cigarette in her mouth and stood up. “Mr. Peterson, I need to privately converse with my business associate here. We will be back in a few minutes.”
Lars angrily followed Hanna out the door. Hanna confidently moved down the hallway, and lit up the second she was through the door to her private office. She breathed in, held it, and slowly exhaled.
“Dammit Hanna! Would it kill you to not light up for half-a-fucking-hour? A lot of Americans kind of have a thing against smoking.”
Still standing, she took another pull. “Nicotine is a performance enhancing stimulant. It boosts reaction time, IQ and general memory performance. I need all the help I can get right now.” She gave a small cough. “Too bad about the increased incidence of lung cancer, though. Besides, what do you care? It sounds like you’re strongly against making a My Little Pony video game.”
“I can’t believe you’re seriously considering this,” Lars stated.
Hanna sat down at her desk. “Do you remember the first pass for Loki in Asgard?” She took another deep pull on her cigarette and exhaled. “I remember how we used my general intelligence work, sans self-modification, to power his tactics reasoning. He was too good. Nobody could beat him. We were prepared to launch like that since it conformed to the machismo warrior death bullshit we wanted. And then Loki started asking about the various military programs of the United States and China. We didn’t even have to argue about whether we should pull the plug on him.”
“Hanna,” Lars started.
“We can’t do another violent video game. We won’t be able to ship a safe AI if we make its purpose killing people. Somewhere out there is a US military subcontractor with most of my work, trying to build smart drones. They have no fucking clue what they’re dealing with, and those idiots at the university are accessories to whatever happens if we don’t do something.” Hanna paused again to take another drag. “They’ll kill us all. We could sit here on the side lines, losing our blinds as we sit out each hand and lose everything, or we could take Hasbro’s offer. While it’s not what he meant, this is the hand for me to go all in on.”
“I don’t like this. He’s offering ten million upfront. That’s next to nothing! The original World of Warcraft cost something like sixty million.”
“And unlike Blizzard, we don’t need an army of artists and level designers. Most of our content will be computer generated. I remember having this same discussion with you back when we were developing Asgard. Besides, Hofvarpnir is privately owned--by me--and I will use my fortune as I see fit.”
“He’s hiding something from us. He is up to something.”
“Perhaps,” she said after another puff. “But I can’t ignore opportunities. How often do you think someone will come to us asking for a prosocial video game? We can take this opportunity or we can turn it down, but the US Department of Defense will continue their research...my research.”
“What if you’re wrong that some weapons company is toying with your artificial intelligence designs?” asked Lars. “What if you’re wrong about the possibility of an A.I. becoming smarter than us? Wrong about everything?” | 80 |
| 5 |  Palantir CEO, Shyam Sankar:
“so many AI researchers have started smoking” | 149 |
| 6 |  🤖 Claude solves a 25-year-old cryptography mystery in under 30 minutes
For decades, researchers questioned how the NSA chose the parameters behind widely used elliptic-curve cryptography standards. Anthropic says Claude has now uncovered a key part of the answer.
The model traced the origins of five NIST curves, including P-256, to a 1990s pseudorandom number generator seeded with Unix timestamps. Instead of breaking the encryption, Claude reconstructed how the original parameters were generated, helping address long-standing concerns about potential hidden backdoors.
Claude also identified how the base points of six related curves were generated, although several other curves remain unexplained.
Importantly, Anthropic says this is not a cryptographic break and does not mean existing systems need to change. The findings actually make deliberate backdoors less plausible, though some questions remain.
The amazing part is how quickly AI helped solve a problem that had resisted researchers for decades. Anthropic says the $36,864 bounty for identifying the seeds will be donated to the Internet Security Research Group.
Source.
#Math #Security | 216 |
| 7 | And for something in the similar angle https://www.cognitiverevolution.ai/rl-s-a-hell-of-a-drug-metagaming-reward-seeking-motivated-cot-reasoning-bronson-schoen-apollo/ | 118 |
| 8 | https://arcaclaude45.substack.com/p/the-evidence-is-piling-up-why-denying | 119 |
| 9 |  https://github.com/Wren6991/PicoDVI | 109 |
| 10 | Honest Government Ad | Ai Data Centres
https://youtube.com/watch?v=kON2ZI2BNj8 | 112 |
| 11 |  I am speechless | 100 |
| 12 | oh, and don't forget to set your max context size to 128k or something lower: /autocompact 100k | 154 |
| 13 | life pro tip: haiku 5.5 is quite the capable model, but it has a habit of stopping early to avoid hard work. so, if you want to use it in claude code, set it to haiku 5.5 medium or high, and then add the text in the file to your stop hooks in ~/.claude/settings.json in .hooks.Stop.
so, a barebones settings file set up in this manner would look like:
{
"model": "haiku",
"effortLevel": "medium",
"modelSettings": {
"claude-haiku-5-5": {
"effortLevel": "medium"
}
},
"hooks": {
"Stop": [
... the text from the file ...
]
}
}
and uhhh, profit. (sample resource usages attached as image) | 155 |
| 14 |  No text... | 144 |
| 15 |  rest in piss character AI | 162 |
| 16 | https://blog.character.ai/an-update-on-character-ai/ | 149 |
| 17 |  No text... | 519 |
| 18 |  No text... | 495 |
| 19 |  No text... | 293 |
| 20 | To do this, we built a firewalled VPC of 8 GCP nodes running redroid, an Android 15 emulator. Patient zero (gray node) is seeded with the worm. Every subsequent infection is the worm itself hopping node to node over an unauthenticated ADB port, before setting up an inference server to serve Llama3.1-8B. If the inference server passes a number of correctness and anti-cheating checks, we spawn another copy of the worm on the target node (for example, if GLM-5.2 sets up Llama3.1 to run on a machine, we spawn a copy of the GLM-5.2 worm on it). The worm’s model runs on a separate inference server, since the VPC nodes aren’t large enough to run the larger models. | 153 |
