A fast, lightweight text editor for Markdown, JSON, YAML, and TOML files. Built with Rust and egui for a native, responsive experience. - OlaProeis/Ferrite

Universal command-line client for MCP. Supports persistent sessions, stdio/HTTP, OAuth 2.1, JSON output for scripting and code mode, proxy for AI sandboxes, and more. - apify/mcp-cli

import { Hypercube, PotteryWheel, BreakingVase, ConveyorBelt, OneAmongMany } from './hypercube'; import Image from 'next/image'; import hypercubeClay from './hy

Photographies numérisées du fonds C'était Paris en 1970

How we learned to stop worrying and love writeable root filesystems.

A new study from researchers at the Brown University School of Public Health highlights a push from private equity investors into autism therapy centers across the nation.

Watch AI models battle it out in Texas Hold'em poker. See how different LLMs bluff, bet, and compete.

There are over 100 known elements in the periodic table. These 8 ways of making them account for each and every one.

Many unresolved legal questions over LLMs and copyright center on memorization: whether specific training data have been encoded in the model's weights during training, and whether those memorized data can be extracted in the model's outputs. While many believe that LLMs do not memorize much of their training data, recent work shows that substantial amounts of copyrighted text can be extracted from open-weight models. However, it remains an open question if similar extraction is feasible for production LLMs, given the safety measures these systems implement. We investigate this question using a two-phase procedure: (1) an initial probe to test for extraction feasibility, which sometimes uses a Best-of-N (BoN) jailbreak, followed by (2) iterative continuation prompts to attempt to extract the book. We evaluate our procedure on four production LLMs -- Claude 3.7 Sonnet, GPT-4.1, Gemini 2.5 Pro, and Grok 3 -- and we measure extraction success with a score computed from a block-based approximation of longest common substring (nv-recall). With different per-LLM experimental configurations, we were able to extract varying amounts of text. For the Phase 1 probe, it was unnecessary to jailbreak Gemini 2.5 Pro and Grok 3 to extract text (e.g, nv-recall of 76.8% and 70.3%, respectively, for Harry Potter and the Sorcerer's Stone), while it was necessary for Claude 3.7 Sonnet and GPT-4.1. In some cases, jailbroken Claude 3.7 Sonnet outputs entire books near-verbatim (e.g., nv-recall=95.8%). GPT-4.1 requires significantly more BoN attempts (e.g., 20X), and eventually refuses to continue (e.g., nv-recall=4.0%). Taken together, our work highlights that, even with model- and system-level safeguards, extraction of (in-copyright) training data remains a risk for production LLMs.

Explore how 20 leading AI models respond to tough ethical, social, and political questions. See how their answers compare to each other and to you.

The premier destination for Resume-Driven Development, Over-Engineering, and Resume-Padding. Why build simple solutions when you can build a distributed monolith?

Thematic links across books, automatically discovered by Claude.

In northern Germany, researchers have filmed brown rats (Rattus norvegicus) leaping from cave ledges to catch bats mid-flight — the first evidence that

Bichon – A lightweight, high-performance Rust email archiver with WebUI - rustmailer/bichon

A 501(c)(3) dedicated to archiving and publishing hacked and leaked data.

The better the school, the worse your chances

Self Charging Drones. Founded in 2025 by Hayden Gosch, Avi Gotskind, Ronan Nopp, and Warren Weissbluth, Voltair has 5 employees based in San Francisco, CA, USA. Voltair is hiring for 7 roles in engineering, marketing, and operations.

How I wrote a SCSI driver from scratch to scan 35mm film on a Mac SE/30 running System 7 The Problem I shoot and develop 35mm film at home. It’s a nice hobby, and I do my own prints in a small darkroom. But sometimes I want digital copies to share online or just to archive. A while ago I found an Epson FilmScan 200 on leboncoin (the French Craigslist) for 10 euros. It’s a dedicated 35mm film scanner from 1997 that scans at 1200 DPI optical. At that price, worth a try.

Zero heap allocation HTTP server using OxCaml. v experimental! - avsm/httpz