ChatGPT starts blocking direct requests to copy an author's style
New behavior capturing a writer's "broad qualities" could have legal implications.
→ View original source
Show HN: FeyNoBg – Automatic background removal model and training library
FeyNoBg is an automatic background removal model and training library presented on Hacker News. It is designed for programmatic background segmentation and appears to include tooling for model training. The project was s…
→ View original source
Kimi K3 Under the Hood: A 2.8T MoE Built for Million-Token Agents
Kimi K3 is a native multimodal Mixture-of-Experts model released by Moonshot AI with 2.8 trillion total parameters, 104 billion active parameters, and a one‑million‑token context window. It incorporates hybrid attention,…
→ View original source
Nvidia CEO Jensen Huang defends Open Source AI by saying distillation is fundamental to learning
Nvidia CEO Jensen Huang argues that distillation—learning from AI, people, and other knowledge sources—is essential for intelligence and should not be viewed as theft. He emphasizes that AI must continually acquire knowl…
→ View original source
Elevated errors on Claude Opus 5
An incident affecting Claude Opus 5, a large language model developed by Anthropic, has been reported on the official status page. The issue involves elevated error rates impacting model performance. Users may experience…
→ View original source
AI watermarking: a green/red token bias seeded by the prior word, a z-test detector, and why paraphrasing breaks it
The question I keep getting asked is "how do you prove a piece of text was written by a machine — without access to the machine?" AI watermarking is the surprisingly clever answer, and once I understo
→ View original source
NVIDIA just announced Open Secure AI Alliance with goal to build and share open tools that promote responsible use of and trust in AI
NVIDIA has launched the Open Secure AI Alliance, a collaborative initiative involving approximately 35 partners including Microsoft, IBM, and Hugging Face. The alliance aims to develop and share open models, harnesses, a…
→ View original source
Kimi-K3 Releases on HuggingFace 7/27
Moonshot AI has released the Kimi-K3 model on HuggingFace. This release provides access to the model's weights and architecture via the official repository. Read original
→ View original source
Claude Opus 5 Has Arrived — Here’s Why Anthropic’s New AI Model Is Turning Heads
Anthropic has released Claude Opus 5, a new AI model designed for the era of AI agents. The model is engineered to be faster, smarter, and more affordable, signaling a new direction for enterprise AI applications. Read o…
→ View original source
Meta has confirmed that it will release an open source model in the future
Meta has confirmed its intention to release a new open source model in the future. This announcement was shared via social media and discussed within the LocalLLaMA community. Read original
→ View original source
The "LLM API" Market Isn't One Market Anymore — It's Splitting Into Tiers, and Qwen Is a Good Case Study
The LLM API market is fragmenting into distinct tiers rather
→ View original sourceHigh-Performance C++20 Optical Neural Network (ONN) Simulator
A new C++20 simulator for Optical Neural Networks (ONNs) delivers high‑performance hardware‑level acceleration by modeling matrix multiplication via light interference in Mach‑Zehnder Interferometer (MZI) meshes. The too…
→ View original source