are you having fun in the singularity?

Joined October 2021
More of this!
Introducing Bitwater: the water heater element that mines Bitcoin. It screws into the electric tank you already have. Physically decentralized mining that runs no matter what 🧵 Heat water. Mine Bitcoin. Same tank.
44
I think @bot going to win, you guys. I just find myself returning to it every day.
58
Everyone's hyped on FastH3 Omni, so I ran it on my 5090 against the Fast H3 I already use. Same shot, warm runs: - Fast H3 in ComfyUI: 44s at 1344x768 - Omni in FastVideo: 74s, and only 832x480 because 768p wouldn't fit - Omni in ComfyUI (NVFP4): 118s at 768p So I'm sticking with Fast H3 for real work right now. Anyone getting better Omni numbers on a 5090? I'd love to know what I'm missing.
1
107
I think I have a good analogy for what's about to happen in cyber. It's pitch dark. Always has been. And the world is dilapidated buildings and rats and maggots and other creepy-crawlies. The whole world. And we just invented flashlights.
55
119
14
1,580
62,061
PSA: if you've ever had a post pop off on here, go check your message requests. My OpenClaw / Moltbook post blew up back in February. I just found like 13 DMs from that week sitting in the requests folder. Had no idea that folder existed 🤦‍♂️ Sorry to anyone I ghosted. (@bradmillscan ) Some were real people reaching out. Check it. I've got Grokbot checking mine every morning now. 😅
1
183
How many of these can u guys deploy to Paris?
Unitree Introducing: H2-X Kung Fu Legend Guardian🤖 Powerful performance to serve and guard you, with a maximum endurance of approximately 8 hours. Please use the robot safely and in a friendly manner.
172
Hell yes. Local video gen just got faster. Will be putting the 5090 through its paces tonight.
As promised, FastVideo open-sources @MiniMax_AI FastH3 Omni, training code included! 🚀 - Real-time gen @ 12x faster on NVIDIA Blackwell - NVFP4/FP8 for Spark + RTX! - ComfyUI day-0 Collab with @NVIDIAAI , Nuva Lab @nuvalab and Reactor @reactorworld Open source must win! ♥️
141
I foresee no problems
claude code just reverse-engineered my hotel room's door lock in 90 seconds. the encryption is from 2014. the same protocol controls the smoke detector in the top right corner - that red light isn't a battery indicator. it blinked twice when i ran the scan. i'm lying in a $85/night hotel room right now. opened claude code and typed: "scan every bluetooth and wireless device in range. show me what's talking to what." it found 19 wireless devices. 18 were normal. device 19: a BLE controller running an unpatched firmware from 2014. it manages the electronic door lock, the smoke detector, the thermostat, and the room safe. all four on one controller. one protocol. zero updates in 10 years. i asked claude code to analyze the protocol. it took 90 seconds. the lock uses a 4-byte rolling key with no session tokens. claude code generated a valid master key on the third attempt. i heard my door click from the bed. then it found the smoke detector is on the same bus. the red light in the photo? that's the controller's status LED. it's not monitoring smoke. it's confirming the BLE handshake with the lock system every 8 seconds. that means anyone within bluetooth range - the hallway, the room next door, the parking lot - with one prompt and 90 seconds can: 1. unlock your door 2. disable the room alarm 3. open your safe 4. and the smoke detector confirms the connection with that little red blink the check takes 2 minutes: step 1 → open claude code → paste the prompt from the first comment step 2 → it scans every BLE device in range and flags unpatched firmware step 3 → if it finds one → request a room change → ask for a room with a physical deadbolt → and use it every hotel chain running ASSA ABLOY or Onity locks from before 2018 has this. that's roughly 30% of hotels worldwide. i tested 3 hotels this month. all 3 vulnerable. look at the photo. top right. that red dot blinking every 8 seconds. that's the handshake. save the prompt. next hotel you check into → run it before you unpack.
Readers added context they thought people might want to know
The post claims an AI reverse-engineered a hotel door lock's 2014 BLE protocol also controlling the smoke detector, but documented vulnerabilities required physical port access and smoke detectors are independent. arstechnica.com/security/2012/… github.com/lolptdr/onity-… digitaltrends.com/computing/how-… x.com/0xGoldTooth/st…
124
⚡️Mirthtime ⚡️ retweeted
One consequence of AI has been to induce an ongoing sense of anxiety and agitation in smart, high agency people - that they are not doing enough with this genius tech, embarrassment that they are unable to get the supposedly easy $1000 per month out of their computer.
46
66
15
2,307
67,689
Huge alpha in learning when you don't wanna work with someone and cutting ties early. Wow. So much time and pain saved.
96
What sorcery is this
Meet Nigel. He’s not enjoying our cause-and-effect demo. So I let him sit in the DeLorean. Still not happy. I didn’t write his lines. What he says is generated from his personality and the context of the shot. It’s an entirely new way to create: place imaginary characters in your world, record your performance, and let Simulon generate their reactions and realistically integrate them into your footage. Reality-Aware Rendering is coming to @Simulon this week.
3
327
Okay I’m switching to @sunra_ai for this month to see if it can save me big bucks on commercially licensed video gen for clients. Seems waaaay cheaper than using Fal or Higgsfield for models like seedance 2.5
Today we're launching Sunra MCP, with the most accessible pricing in the industry. Every leading image, video and audio model, right inside Claude, ChatGPT and the AI assistant you already use. Each one at its maker's own API price, with no markup added. Now is the time to build. Go create something no one has seen before.
1
2
246
⚡️Mirthtime ⚡️ retweeted
#Bitcoin has changed my relationship to the physical world. I used to get frustrated when trying to grab my coat that’s stuck in the closet. Now I just think, the very same physics that is causing my coat to be stuck is securing my wealth, and all frustration goes away.
1
1
119
I ran the new MOSS-Image (open 5B from @MosiAI_Official / @Open_MOSS) vs @Alibaba_Qwen's Qwen-Image 2.1 on my 5090. 18 prompts: text-heavy design plus straight photo and art. On text: 3 misspellings for MOSS, about 13 for Qwen. First image is always MOSS, second is Qwen. 1/22
1
343
21/22 My take: MOSS wins narrowly when text and layout have to be right, and it looks more real on people and candid photos. Qwen stays my pick for polish, landscapes, anime and fast drafts. Keeping both. Will likely A/B with both for the next few weeks any time I need image Gen, and then settle on a favorite. Anyone tried MOSS's JSON mode?
1
126
22/22 Links if you want to try it: MOSS-Image-SFT (Apache-2.0): huggingface.co/OpenMOSS-Team… Code: github.com/OpenMOSS/MOSS-Ima… ostris on the training data: xitter.cf/ostrisai/status/211159…
I think one of the reasons people love MOSS Image so much is they intentionally removed AI images from their training data. There is a clear trend of new models training on AI images and having an overly ChatGPT look to them. Moss just looks better because they didn't do that.
106