Communick News

cross-posted from: https://feddit.org/post/36946211

https://hexane.cc/

AFD email accounts were hacked:

• 421,179 emails

• Photo IDs

• Financial records

• Personal documents

• Member lists

Crossposted from https://lemmy.world/post/53323419

cross-posted from: https://piefed.world/c/tech/p/1476563/greg-abbott-raked-in-9-million-from-data-center-donors

cross-posted from: https://lemmy.ca/post/72623197

There's an app being presented in a small store with Linux apps, without any licensing information or link to its source code. The package archive contains mostly scripted code, one compiled executable, and this license:

Copyright (c) 2026 .... All rights reserved.

Proprietary software.

Third-party components and assets retain their respective licenses and notices.

There is no additional licenses at all in the whole package, not even in the SVGs that the dev most definitely did not create from scratch.

The app probably pulls in these third party components during runtime, or provides them inside the archive. Short of analyzing every single file in the package I have no idea.

Most of these third party components seem to be pulled from the developers own website (murky), one or two are not, and those are most definitely open-source-licensed, probably GPL.

I raised an issue about transparency requirements in the app store. Not about this one developer - yet.

Most contributors do provide source code and a GPL or similar license, but the app store's policy phrases this as a nice thing, not a must. But there's always been a strong faction that does not care about such things at all. And there's a rising tide of new, mostly LLM-assisted apps that have no concept of transparency, or qualms about stealing code - be it through the LLM or directly.

All in all this is one of those projects that mix FOSS-licensed Linux stuff with proprietary-licensed and closed source stuff in a very unfortunate way.

Is there any sort of legal situation here?

Apart from that - what else would you say about murky ethics here? What arguments can I use to make my point? There's many similar packages up on that store.

Many people suspect that this developer is lying when they say they're not using LLMs for development. Any way to disprove that claim based on the package archive?

While writing this I analyzed the software package in question a bit more and I'm pretty sure I know the answer already, but I'm still asking the question. Maybe I missed something, or maybe there is a legal issue I'm not aware of...

Disclaimer: if you think I posted this in the wrong community please refrain from commenting or downvoting, use the report button instead. Thank you. FWIW, I did read the sidebar and imo it is open-ended enough to allow for questions like this one.

transcription: someone in the street asked me "would you kiss a girl for 1000$" and i started to cry bc i cant pay that much

I've been working with MOSS-Image, the open text-to-image model from the OpenMOSS team (Apache-2.0 weights), and wanted to write down the settings and prompt habits that matter most, since most "settings guides" for image models are vague.

Quick model facts that matter for prompting

• It's a ~5B-parameter diffusion transformer with a Qwen3.5 text encoder and the FLUX.2 VAE. The LLM-class encoder is the reason long, structured prompts are followed better than with CLIP-style stacks.
• There are two checkpoints: Base (for LoRA training / fine-tuning) and SFT (instruction-tuned). For direct prompting, use SFT. Base gives noticeably looser prompt adherence.

Settings worth understanding

1. Steps. The recommended range is 50+. Small text and fine textures are the first things to degrade when you cut steps to save time.
2. Guidance. 3.0-6.0 is the useful range. The low end is more natural and varied; the high end follows the prompt more literally but can get contrasty. Start in the middle and move in small increments.
3. Seed. Fix it while you compare anything else. Changing seed and guidance at the same time makes it impossible to tell what helped.
4. Sampler. Euler is the default and a sensible baseline.
5. Negative prompt. Short and specific ("blurry text, extra letters, watermark") tends to beat long boilerplate lists.

Prompt structure

• Put any text that should appear in the image in quotes, exactly as written, and say where it goes ("title at the top", "small date at the bottom").
• Keep each text block short. Two or three short strings render far more reliably than one sentence.
• For layouts, describing the composition as a small JSON object (canvas, background, elements with positions, style) tends to be followed more reliably than one long sentence. Again, that's mostly the encoder.

Running it without a local GPU

Local setup needs a lot of VRAM for 2K output. Disclosure: I'm on the team behind Moss Image, an independent hosted front-end for this model (not affiliated with OpenMOSS). New accounts get free credits, so you can test the settings above in a browser: Visit Site

Curious whether anyone has compared it with other open models specifically on text rendering, which is where open models seem to differ most.

A lot of "flying car" clips have been going around lately, and a fair number of people in my feed thought some of them were real. If you make these, here are a few things that help with getting a result that still looks like your own car, and about sharing it honestly.

Getting a clip that keeps the car's identity

• One sharp photo with the whole car in frame. Every wheel visible, no cropped bumper. Anything the model can't see, it invents, and invented parts are where the car starts morphing.
• Three-quarter or side angle. A front three-quarter view carries the most information about the car's shape.
• Uncluttered background. A parked car on an empty lot animates better than one squeezed between other cars.
• One motion per clip. A lift-off, a hover or a slow fly-by. Combining several moves in a short clip is where consistency breaks.
• Short and vertical. Around 5-10 seconds in 9:16 suits Reels/Shorts and leaves less time for drift.
• Check wheels, badges and plates at full size before posting. Those small details show artifacts first.

Labelling

Put "AI-generated" in the caption or as an on-screen tag. These clips look convincing on a phone, and an honest label keeps the comments about the edit instead of an argument about whether it's real. Some platforms also have their own AI-content labels now. Use them.

Disclosure: I work on Flying Car AI, a template-based photo-to-video tool for exactly this kind of clip. It's independent and not affiliated with any car maker: Learn More

Happy to hear what's worked for others, especially for keeping wheel shapes stable during the motion.

• System : archlinux
• WM : river-classic
• Bar : yambar
• Colorscheme : nord
• Terminal : kitty
• Editor : doom emacs
• Wallpaper : found in there

In this powerful keynote, researcher Joshua Doss argues that we cannot dismantle and replace the current economic system without explicitly naming and tackling the systemic racism and inequality that have become inseparable from the system of capitalism.

Working-class Americans across all backgrounds and races are being "hunted" by the same extractive economic monster. The historical role of Black exploitation that made American capitalism possible is starting to be experienced to some degree by all of the working class. Will this lead to a class consciousness of the working class?

Such internet restrictions are more commonly associated with countries like China, Iran, Russia and, more recently, India.