CTO - Agentic AI / Computer Graphics Engineer youtube.com/@NabilTechArt Unlocking the Singularity in Software & Computer Graphics apeironengine.com
Carthage
- Tweets 589
- Following 71
- Followers 33
- Likes 517
xD
Introducing Unity Spark — a new way to create a game, iterate on it with others, publish it, and play it - all without writing a line of code.
You'll build with artist-created assets from the Unity Asset Store, and then you can take your game to the next level by refining it in a new web-based editor.
Prompt it. Perfect it. Publish it. Inspired games, with artist-created assets, built on Unity.
Closed beta coming soon.
6
Nabil KH retweeted
the starting template for @kaigen3d is 400 lines of C code and gives you:
- a physics based character controller
- procedural skybox supporting day night cycles
- customizable editor
- runs natively on PC, mobile and web
you can point your AI agent and go nuts
10
8
216
4,216
Amazing choice for mold to go with rust
A Linux developer tool just dropped its entire C++ codebase for Rust.
It’s called mold, and it helps developers build software faster by combining compiled code into the final program.
Version 3.0 is a complete Rust rewrite, and the developers say it’s still just as fast as the C++ version.
It keeps the same commands and supported architectures. They even built all Gentoo packages to check compatibility and reported no regressions.
The rewrite also replaces CMake with Cargo, removes Intel’s oneTBB dependency and adds bounds checking when processing corrupted files.
Now the developers want to close the remaining compatibility gaps with GNU ld so Linux distributions can adopt mold as their default.
Source: github.com/rui314/mold/relea…
6
this is great and very needed against the uncensored models , but its very important defensive capabilities reach the most vulnerable too ( individuals & small companies ) maybe via a product built on top of Anthropic Authorized capabilities ? that way its much easier for @AnthropicAI to make sure its product is being used safely
We’re expanding our Cyber Verification Program to give security professionals broader access to our most capable models.
Through this program, verified security professionals can access Claude Mythos 5.1, Opus 5.5, and Sonnet 5.5 with safeguards designed for defensive work.
We’re also opening up new tiers to allow for authorized offensive work, like penetration testing and red-teaming.
anthropic.com/news/cyber-ver…
10
Nabil KH retweeted
🚨JUST IN: OpenAI models will now have invisible watermarks hidden in ChatGPT and Codex outputs
>"We are not making text watermarking a global default AT LAUNCH"
>"phased approach starting with the EU"
>"this regional approach gives us room to learn"
>"starting today, API customers around the world WILL be able to opt in…"
"We expect to revisit each part of this approach as the technology, standards, and evidence evolve"
26
420
7,983
424,067
this render any paid software that runs on the user machine ( including webapps that are entirely wasm based ) literally free to change as you wish ( almost ) , so we will have to come up with a new approach how we ship and consume software in this era . ( those that runs core product functions on web backends etc , probably they are safe but not everyone have the luxury to run valuable product features on the cloud )
YOUR LLM CAN NOW REVERSE ENGINEER BINARIES ON ITS OWN 💀
ghidraMCP connects Ghidra to MCP clients
◾️ decompiles and analyzes binaries
◾️ renames methods and data
◾️ explores methods, classes, imports & exports
◾️ works with Claude Desktop, Cline & 5ire
7
Technically is really really hard to stop distiliation even by stripping away the thinking tokens ? you could always use a SOTA open weight model and alongside humans it can generate the thinking tokens for that output , and you could automate that so you could extract significant data at scale ... maybe the SOTA AI models can go behind application layers instead ?
This explains a lot about why Anthropic hates Chinese models and open source.
Anthropic’s Nick Marwell says frontier AI could eventually require $10B, $100B, even $1 trillion training runs.
His argument: if a company spends that much pushing the frontier, competitors shouldn’t be able to distill the model and ship a cheap copy weeks later.
That’s the real fight.The problem is that not every Chinese model is just distillation. Chinese labs are spending serious money on training, building their own techniques, and getting extremely good at optimization and efficiency.
And as models keep improving, I think capabilities will increasingly converge. The gap between the top labs may become much smaller, which makes protecting one company’s frontier advantage much harder.
1
12
