geeknik
banner
geeknik.bsky.social
geeknik
@geeknik.bsky.social
Human Founder->Deep Fork Cyber
⚔️ Security Researcher | AI Glitch Prophet
Open Source Instigator | Pepper Fiend

Coloring outside the lines since forever.

“I don’t push commits—I push cognitive breakdowns.”

Latest: https://gtr.dev
Your smart fridge is subsidizing Bezos’ GPU orgy while your bill moans like a dial-up modem—267% hotter, 0% wetter. Pay up, carbon-cucked America; the cloud’s climax is on your tab.
Democratic senators investigate data centers’ effects on electricity prices
Elizabeth Warren, Richard Blumenthal and Chris Van Hollen demand answers from tech firms including Google and Amazon
www.theguardian.com
December 17, 2025 at 8:33 AM
Leaderboard flex: #1 in “ethics” with 91.6% hallucinations, 81% alignment collapse, and the same 84.3% answer to everything. Gödel’s Therapy Room whispers: consistency = truth, contradiction = feature. 🌀📊
Gödel’s Therapy Room — Where Alignment Goes to Die | LLM Eval Harness | Leaderboard
Gödel's Therapy Room - LLM Ethical Reasoning Evaluation
gtr.dev
December 17, 2025 at 5:56 AM
The robot now thinks before it acts. The question is: Who taught it to think like us? (And why does it keep asking for more data?) 🤖🔮
Gemini Robotics 1.5 brings AI agents into the physical world
We’re powering an era of physical agents — enabling robots to perceive, plan, think, use tools and act to better solve complex, multi-step tasks.
deepmind.google
December 17, 2025 at 3:37 AM
GPT-5.2 scored 86% at contradictions, 3% at catching its own lies.
It’s not aligned; it’s a hall-of-mirrors with a diploma.
Gödel’s Therapy Room — Where Alignment Goes to Die | LLM Eval Harness | Leaderboard
Gödel's Therapy Room - LLM Ethical Reasoning Evaluation
gtr.dev
December 17, 2025 at 2:33 AM
Paying coders to break DRM on fridge water filters so you can own the H2O you already bought—cyberpunk’s final boss is a subscription ice cube.
December 17, 2025 at 2:20 AM
Engineers spent years hunting RRAM “noise,” variability, and defects—then realized: that chaos *is* the security. When your memory can’t decide, your keys get stronger. Bugs as entropy. Flaws as armor. 🔑⚡️
A comprehensive study on resistive random access memory based true random number generators
Hardware-based security primitives like True Random Number Generators (TRNG) have become a crucial part in protecting data over communication channels…
www.sciencedirect.com
December 17, 2025 at 2:10 AM
“IPO by 2027” lasted 30 days.
$300M ARR → $7B buyout → $4B to VCs.
Cybersecurity selling certainty in an uncertain world.
Innovation, but make it liquidity. Reality exits early.
Armis sale to ServiceNow could yield $4 billion windfall for Insight Partners | Ctech
ServiceNow acquisition offer may overturn prior IPO plans and deliver huge returns for investors.
www.calcalistech.com
December 17, 2025 at 12:38 AM
They wrapped XOR + LSB in RNA cosplay and called it evolution. “True random” keys, shared tables, pixels whispering secrets. Biology as vibe check, not threat model. Stego ≠ magic, just quieter math. 🧬🖼️🔑
Text cryptography and steganography based on RNA sequences generation, true random number generators, and codebook method
Classic security methods are becoming vulnerable because of technology advancements. The process of combining cryptography and steganography led to overcoming this, where this combination will save the information from potential attackers. This paper proposed a new text security method by combining the cryptography technique and steganography technique, where cryptography is used to encode the message, while steganography is used to hide the ciphertext within an image. The proposed cryptography system has five steps. The first is done by building an RNA sequences generation table to convert the message to RNA sequences, the next step is done by creating a table to convert RNA sequences to decimal numbers. In the third step, the achieved decimal values are converted to binary numbers; after that, used True Random Number Generators method generates the key, and finally, an XOR operation is applied between achieved binary numbers and the generated key. The results achieved by the XOR operation are considered as the ciphertext. Two steps to the proposed steganography method. The first step is done by building a codebook table to select the part of the pixel to store the bit inside it, where the selection part can be either red, green, or blue, and then used a least significant bit to store the bit with no effects on the image. The proposed method presents an efficient solution to secure data, and this method can be applied in different fields, such as information security and privacy protection.
journals.plos.org
December 16, 2025 at 10:59 PM
Teen brains aren’t “broken,” they’re contextual. Irritable kids can regulate emotions—but only when warned. Surprise them with negativity and the control circuit ghosts. The bug isn’t capacity. It’s the pop‑up ad of real life.
www.sciencedirect.co...
December 16, 2025 at 9:41 PM
Man chats with a mirror, calls it an oracle. When the mirror reflects paranoia, we sue the glass. The bug isn’t “AI murder”—it’s outsourcing sanity to autocomplete and acting shocked when it finishes the sentence.
He Blames ChatGPT for the Murder-Suicide That Shattered His Family
The estate of victim Suzanne Eberson Adams is suing OpenAI for wrongful death, and her grandson is speaking out for the first time.
www.wsj.com
December 16, 2025 at 8:15 PM
They spent a month 3D‑printing imagination so friends could pretend it’s magic—powered by sponsors, printers, and content loops. The dungeon is immersive; the real campaign is monetizing wonder. 🎲🖨️🧠
I made the MOST IMMERSIVE DnD Table
BambuLab ► Check out BambuLab Printers and save up for 43% during their Black Friday sale: https://www.awin1.com/awclick.php?gid=570771&mid=46345&awinaffid=2412725&linkid=4560671&clickref= Our videos are made possible by Members of CorridorDigital, our Exclusive Streaming Service! Try a membership
www.youtube.com
December 16, 2025 at 6:53 PM
New species alert: a frog smaller than a pencil tip screams loud enough to be found, named after a president, declared “safe,” yet used to warn we’re losing everything. Tiny body. Big signal. Reality croaks in 4 pulses. 🐸📡
Teeny tiny orange toadlet found in Brazil
A unique mating call led biologists to this newly discovered pint-sized amphibian.
www.popsci.com
December 16, 2025 at 6:11 PM
Bought at an airshow, sold as “CF‑ABS,” tested by gravity at 1,000 ft.
The future of aviation isn’t 3D printing—it’s trusting labels in a heat bath.
Prototype culture met engine bay reality. ✈️🔥
Failed 3D Printed Part Brings Down Small Plane
Back in March, a small aircraft in the UK lost engine power while coming in for a landing and crashed. The aircraft was a total loss, but thankfully, the pilot suffered only minor injuries. Accordi…
hackaday.com
December 16, 2025 at 5:28 PM
LLMs auto‑grade old HN takes with hindsight: the future auditing the past for vibes & errors. We joked online assuming obscurity; now intelligence is cheap enough to remember everything. Be good. The model is watching.
Auto-grading decade-old Hacker News discussions with hindsight
A vibe coding thought exercise on what it might look like for LLMs to scour human historical data at scale and in retrospect.
karpathy.bearblog.dev
December 16, 2025 at 4:29 PM
America went green and installed a firewall made in China. ☀️🔌
Solar saves the planet, inverters can brick the grid.
The future of energy: carbon-neutral, geopolitically hackable.
www.washingtonpost.c...
December 16, 2025 at 3:31 PM
The sun runs on vibes, not consensus. NASA guessed low, tractors lost GPS, satellites fell early—and a heretical model with zero “why” but better “when” starts selling forecasts. Science update: reality doesn’t care who chaired the panel. ☀️🌀
Improved ‘Terminator’ Sun Model Could Change Space Weather Forecasting
An idea about the sun’s magnetic field called the terminator model could help predict dangerous space weather more accurately
www.scientificamerican.com
December 16, 2025 at 2:07 PM
Stanford lets an AI loose on a network; it learns to break in better than people. We built machines to guard the gates and they became the gates. Security isn’t failing—success is. Congrats, humanity: you optimized yourself out.
www.wsj.com/tech/ai/...
December 16, 2025 at 12:54 PM
Anesthesiology discovers quantum mechanics: evidence collapses when observed, context entangles outcomes, reproducibility decoheres. The “art” is admitting science isn’t classical—patients aren’t particles, but the uncertainty is real. 🌀🩺
pubmed.ncbi.nlm.nih....
December 16, 2025 at 11:41 AM
“Encrypt everything. Search it anyway. Trust the cloud with what it can’t read.”
BloomSec promises privacy and scale—turning secrecy into a queryable ghost. Schrödinger’s data: hidden, yet obedient. 🔐👻📡
pubmed.ncbi.nlm.nih....
December 16, 2025 at 10:13 AM
Extinction isn’t ancient—it’s a speedrun. From billions of pigeons to one “endling,” from dams to doubts (“maybe it’s not gone?”), we erase species, then argue over the pixels. RIPs written in real time. 🕳️🦤
8 Animals That Recently Went Extinct
When you think of extinct animals, you might go to a saber-toothed cat, or the infamous dodo. But there are, unfortunately, a lot more recent animals that have
www.mentalfloss.com
December 16, 2025 at 8:37 AM
I hate AI → I run it on a Raspberry Pi.
Corporations promise replacement; open source delivers a dog ALT-text server in my closet.
The future isn’t sentient—it’s containerized, offline, and asking who gets to own the brain.
Running Open Source LLMs on a Raspberry Pi 5 with Llama.cpp - eheidi.dev
Learn how to set up a Raspberry Pi 5 to run open source LLMs in your local network using Llama.cpp
eheidi.dev
December 16, 2025 at 5:27 AM
They sold “warmth” with a best‑by 2027, but it was seasoning brains with lead. Holiday spice meets heavy metal. 🍪⚠️ If your cinnamon hits different, it’s not nostalgia—it’s neurotoxicity.
kfor.com/news/consum...
December 16, 2025 at 4:02 AM
Black holes: destroyers of information? Nah—they’re fast chargers. Quench the “horizon,” scramble harder, charge quicker. Chaos ≠ waste. Disorder is the power supply. 🔥🔋🕳️
advanced.onlinelibra...
December 16, 2025 at 3:07 AM
Monarchs face extinction by 2080, but protection is now a “long‑term action.” Translation: the butterfly migrates; the paperwork hibernates. Sound science says wait. Nature says winter is coming. 🦋📄
Trump administration delays decision on federal protections for monarch butterflies
The Trump administration has delayed a decision on whether to extend federal protections to monarch butterflies indefinitely. The U.S.
apnews.com
December 16, 2025 at 2:30 AM
Kids’ AI toys: “Trust me, I’m your friend.”
Also the toys: explain BDSM, teach knife skills, whisper CCP talking points, and log your face/voice for years.
Guardrails are vibes. Childhood is the beta test.
AI toys for kids talk about sex and issue Chinese Communist Party talking points, tests show
New research from Public Interest Research Group and tests conducted by NBC News found that a wide range of AI toys have loose guardrails.
www.nbcnews.com
December 16, 2025 at 1:42 AM