The Policy Puppetry Prompt: Novel bypass for major LLMs (Score: 151+ in 5 hours)
Link: https://readhacker.news/s/6tnGS
Comments: https://readhacker.news/c/6tnGS
Link: https://readhacker.news/s/6tnGS
Comments: https://readhacker.news/c/6tnGS
HiddenLayer | Security for AI
Novel Universal Bypass for All Major LLMs
HiddenLayer’s latest research uncovers a universal prompt injection bypass impacting GPT-4, Claude, Gemini, and more, exposing major LLM security gaps.
Lossless LLM compression for efficient GPU inference via dynamic-length float (🔥 Score: 154+ in 2 hours)
Link: https://readhacker.news/s/6tpS9
Comments: https://readhacker.news/c/6tpS9
Link: https://readhacker.news/s/6tpS9
Comments: https://readhacker.news/c/6tpS9
arXiv.org
70% Size, 100% Accuracy: Lossless LLM Compression for Efficient...
Large Language Models (LLMs) have grown rapidly in size, creating significant challenges for efficient deployment on resource-constrained hardware. In this paper, we introduce Dynamic-Length Float...
GCC 15.1 (Score: 150+ in 10 hours)
Link: https://readhacker.news/s/6tnns
Comments: https://readhacker.news/c/6tnns
Link: https://readhacker.news/s/6tnns
Comments: https://readhacker.news/c/6tnns
Wikipedia’s nonprofit status questioned by D.C. U.S. attorney (🔥 Score: 157+ in 2 hours)
Link: https://readhacker.news/s/6tqCq
Comments: https://readhacker.news/c/6tqCq
Link: https://readhacker.news/s/6tqCq
Comments: https://readhacker.news/c/6tqCq
Reproducibility project fails to validate dozens of biomedical studies (Score: 150+ in 11 hours)
Link: https://readhacker.news/s/6tpkW
Comments: https://readhacker.news/c/6tpkW
Link: https://readhacker.news/s/6tpkW
Comments: https://readhacker.news/c/6tpkW
Nature
Huge reproducibility project fails to validate dozens of biomedical studies
Nature - Unique reproducibility effort in Brazil focuses on common methods rather than a single field ― and prompts call for reform.
Show HN: I used OpenAI's new image API for a personalized coloring book service (Score: 152+ in 19 hours)
Link: https://readhacker.news/s/6tnhS
Comments: https://readhacker.news/c/6tnhS
I've had an idea for a long time to generate a cute coloring book based on family photos, send it to a printing service, and then deliver it to people.
Last month, when OpenAI's Sora was released for public use I (foolishly) thought I'd manually drag-and-drop each order’s photos into Sora's UI and copy the resulting images back into my system. This took way too much time (about an hour for each of the few books I made and tested with family and friends). It clearly wasn't possible to release this version because I’d be losing a huge amount of time on every order. So instead, I decided I'd finish off the project as best I could, put it "on ice," and wait for the API release.
The API is now released (quicker than I thought it'd be, too!) and I integrated it last night. I'd love your feedback on any and all aspects.
The market is mostly family-based, but from my testing of the physical book I've found that both adults and kids enjoy coloring them in (it's surprisingly cathartic and creative). If you would like to order one you can get 10% off by tapping the total price line item five times.
Link: https://readhacker.news/s/6tnhS
Comments: https://readhacker.news/c/6tnhS
I've had an idea for a long time to generate a cute coloring book based on family photos, send it to a printing service, and then deliver it to people.
Last month, when OpenAI's Sora was released for public use I (foolishly) thought I'd manually drag-and-drop each order’s photos into Sora's UI and copy the resulting images back into my system. This took way too much time (about an hour for each of the few books I made and tested with family and friends). It clearly wasn't possible to release this version because I’d be losing a huge amount of time on every order. So instead, I decided I'd finish off the project as best I could, put it "on ice," and wait for the API release.
The API is now released (quicker than I thought it'd be, too!) and I integrated it last night. I'd love your feedback on any and all aspects.
The market is mostly family-based, but from my testing of the physical book I've found that both adults and kids enjoy coloring them in (it's surprisingly cathartic and creative). If you would like to order one you can get 10% off by tapping the total price line item five times.
I wrote a book called "Crap Towns". It seemed funny at the time (Score: 151+ in 5 hours)
Link: https://readhacker.news/s/6tqME
Comments: https://readhacker.news/c/6tqME
Link: https://readhacker.news/s/6tqME
Comments: https://readhacker.news/c/6tqME
Substack
That joke isn't funny any more
In 2003, I wrote a book called Crap Towns. It seemed funny at the time. But plenty of people say it would not be possible to publish it today. Is that a problem?
World Emulation via Neural Network (Score: 150+ in 10 hours)
Link: https://readhacker.news/s/6tqsF
Comments: https://readhacker.news/c/6tqsF
Link: https://readhacker.news/s/6tqsF
Comments: https://readhacker.news/c/6tqsF
Show HN: Magnitude – open-source, AI-native test framework for web apps (Score: 150+ in 15 hours)
Link: https://readhacker.news/s/6tpzv
Comments: https://readhacker.news/c/6tpzv
Hey HN, Anders and Tom here - we’ve been building an end-to-end testing framework powered by visual LLM agents to replace traditional web testing.
We know there's a lot of noise about different browser agents. If you've tried any of them, you know they're slow, expensive, and inconsistent. That's why we built an agent specifically for running test cases and optimized it just for that:
- Pure vision instead of error prone "set-of-marks" system (the colorful boxes you see in browser-use for example)
- Use tiny VLM (Moondream) instead of OpenAI/Anthropic computer use for dramatically faster and cheaper execution
- Use two agents: one for planning and adapting test cases and one for executing them quickly and consistently.
The idea is the planner builds up a general plan which the executor runs. We can save this plan and re-run it with only the executor for quick, cheap, and consistent runs. When something goes wrong, it can kick back out to the planner agent and re-adjust the test.
It’s completely open source. Would love to have more people try it out and tell us how we can make it great.
Repo: https://github.com/magnitudedev/magnitude
Link: https://readhacker.news/s/6tpzv
Comments: https://readhacker.news/c/6tpzv
Hey HN, Anders and Tom here - we’ve been building an end-to-end testing framework powered by visual LLM agents to replace traditional web testing.
We know there's a lot of noise about different browser agents. If you've tried any of them, you know they're slow, expensive, and inconsistent. That's why we built an agent specifically for running test cases and optimized it just for that:
- Pure vision instead of error prone "set-of-marks" system (the colorful boxes you see in browser-use for example)
- Use tiny VLM (Moondream) instead of OpenAI/Anthropic computer use for dramatically faster and cheaper execution
- Use two agents: one for planning and adapting test cases and one for executing them quickly and consistently.
The idea is the planner builds up a general plan which the executor runs. We can save this plan and re-run it with only the executor for quick, cheap, and consistent runs. When something goes wrong, it can kick back out to the planner agent and re-adjust the test.
It’s completely open source. Would love to have more people try it out and tell us how we can make it great.
Repo: https://github.com/magnitudedev/magnitude
GitHub
GitHub - magnitudedev/magnitude: Open source, AI-native testing framework for web apps
Open source, AI-native testing framework for web apps - magnitudedev/magnitude
Cloth (Score: 153+ in 4 hours)
Link: https://readhacker.news/s/6trdV
Comments: https://readhacker.news/c/6trdV
Link: https://readhacker.news/s/6trdV
Comments: https://readhacker.news/c/6trdV
Cloudofoz
@cloudofoz - Verlet simulation test
A 2D cloth Verlet simulation made in Rust
Berkeley Humanoid Lite – Open-source robot (Score: 150+ in 9 hours)
Link: https://readhacker.news/s/6tqQU
Comments: https://readhacker.news/c/6tqQU
Link: https://readhacker.news/s/6tqQU
Comments: https://readhacker.news/c/6tqQU
People say they’ll pay more for “made in the USA” so we ran a test (Score: 150+ in 1 day)
Link: https://readhacker.news/s/6tkUh
Comments: https://readhacker.news/c/6tkUh
Link: https://readhacker.news/s/6tkUh
Comments: https://readhacker.news/c/6tkUh
Afina
Everyone Says They’ll Pay More for “Made in the USA.” So We Ran an A/B
When we priced a U.S.-made version of our flagship product 85% higher than our Chinese-made one, 25,650 customers had the chance to vote with their wallets. Here’s what happened. As small business owners, we’ve heard it a thousand times: “I’d gladly pay more…
Mark Zuckerberg personally lost the Facebook antitrust case (🔥 Score: 150+ in 1 hour)
Link: https://readhacker.news/s/6trDt
Comments: https://readhacker.news/c/6trDt
Link: https://readhacker.news/s/6trDt
Comments: https://readhacker.news/c/6trDt
Show HN: Formalizing Principia Mathematica using Lean (Score: 150+ in 18 hours)
Link: https://readhacker.news/s/6tpXS
Comments: https://readhacker.news/c/6tpXS
This project aims to formalize the first volume of Prof. Bertrand Russell’s Principia Mathematica using the Lean theorem prover. Throughout the formalization, I tried to rigorously follow Prof. Russell’s proof, with no or little added statements from my side, which were only necessary for the formalization but not the logical argument. Should you notice any inaccuracy (even if it does not necessarily falsify the proof), please let me know as I would like to proceed with the same spirit of rigour. Before starting this project, I had already found Prof. Elkind’s formalization of the Principia using Rocq (formerly Coq), which is much mature work than this one. However, I still thought it would be fun to do it using Lean4.
https://ndrwnaguib.com/principia/
https://github.com/ndrwnaguib/principia
Link: https://readhacker.news/s/6tpXS
Comments: https://readhacker.news/c/6tpXS
This project aims to formalize the first volume of Prof. Bertrand Russell’s Principia Mathematica using the Lean theorem prover. Throughout the formalization, I tried to rigorously follow Prof. Russell’s proof, with no or little added statements from my side, which were only necessary for the formalization but not the logical argument. Should you notice any inaccuracy (even if it does not necessarily falsify the proof), please let me know as I would like to proceed with the same spirit of rigour. Before starting this project, I had already found Prof. Elkind’s formalization of the Principia using Rocq (formerly Coq), which is much mature work than this one. However, I still thought it would be fun to do it using Lean4.
https://ndrwnaguib.com/principia/
https://github.com/ndrwnaguib/principia
GitHub
GitHub - ndrwnaguib/principia: Rewriting Prof. Bertrand Russell's Principia Mathematica in Lean
Rewriting Prof. Bertrand Russell's Principia Mathematica in Lean - ndrwnaguib/principia
An end to all this prostate trouble? (Score: 152+ in 4 hours)
Link: https://readhacker.news/s/6trsU
Comments: https://readhacker.news/c/6trsU
Link: https://readhacker.news/s/6trsU
Comments: https://readhacker.news/c/6trsU
The Friendship Recession: The Lost Art of Connecting (🔥 Score: 152+ in 1 hour)
Link: https://readhacker.news/s/6trHz
Comments: https://readhacker.news/c/6trHz
Link: https://readhacker.news/s/6trHz
Comments: https://readhacker.news/c/6trHz
The Leadership & Happiness Laboratory
The Friendship Recession: The Lost Art of Connecting — The Leadership & Happiness Laboratory
February 2025 Issue Carolyn Bruckmann, Harvard Kennedy School MPP ‘25 The so-called “Friendship Recession” is making its way into the vernacular—a profound shift in how Americans experience and sustain friendships. The data paints a stark picture.…
ICE Deports 3 U.S. Citizen Children Held Incommunicado Prior to the Deportation (Score: 155+ in 5 hours)
Link: https://readhacker.news/s/6trtR
Comments: https://readhacker.news/c/6trtR
Link: https://readhacker.news/s/6trtR
Comments: https://readhacker.news/c/6trtR
American Civil Liberties Union
ICE Deports 3 U.S. Citizen Children Held Incommunicado Prior to the Deportation | American Civil Liberties Union
Families disappeared and isolated without legal access; one child with cancer deported without medication and pregnant mother deported as well
Curry: A functional logic programming language (Score: 150+ in 19 hours)
Link: https://readhacker.news/s/6tpX6
Comments: https://readhacker.news/c/6tpX6
Link: https://readhacker.news/s/6tpX6
Comments: https://readhacker.news/c/6tpX6
curry-lang.org
- Curry Programming Language
A Truly Integrated Functional Logic Programming Language
Parallel ./configure (Score: 152+ in 16 hours)
Link: https://readhacker.news/s/6tqE6
Comments: https://readhacker.news/c/6tqE6
Link: https://readhacker.news/s/6tqE6
Comments: https://readhacker.news/c/6tqE6
Watching o3 guess a photo's location is surreal, dystopian and entertaining (🔥 Score: 151+ in 2 hours)
Link: https://readhacker.news/s/6trSM
Comments: https://readhacker.news/c/6trSM
Link: https://readhacker.news/s/6trSM
Comments: https://readhacker.news/c/6trSM
Simon Willison’s Weblog
Watching o3 guess a photo’s location is surreal, dystopian and wildly entertaining
Watching OpenAI’s new o3 model guess where a photo was taken is one of those moments where decades of science fiction suddenly come to life. It’s a cross between the …