Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
NVIDIA introduced advanced fused MLP kernels for mixture-of-experts (MoE) models, built with the CuTe DSL, delivering 1.3x–2x kernel-level speedups and enabling sync-free MoE execution. The optimization contributed an 8%…
Anthropic's Claude Fable 5 and Mythos 5 were pulled globally three days after launch due to a US export control directive, disrupting enterprise users. Kilo AI highlights alternative models like OpenAI's GPT-5.5, which o…
Anthropic refused U.S. government demands to restrict its Fable 5 AI model over alleged national security risks, leading to export controls and the model's retirement. The standoff escalated after Amazon reported a jailb…
Tensordyne, the AI chip startup formerly known as Recogni, previewed Napier, an AI inference system that uses logarithmic math to replace multiplication with addition for transformer workloads. The company positions Napi…
Former xAI engineer Devin Kim filed a lawsuit in California state court alleging he was wrongfully fired in September 2025 for repeatedly raising safety concerns about the chatbot Grok. The complaint names xAI and parent…
A study by Peec AI analyzing 1,754 prompts and 37,804 AI responses across five sectors found that over 90% of prompt variations share similar meaning, with wording changes impacting brand mentions only past a threshold. …
A commentator argues that current AI interactions, with their word-by-word streaming responses, mirror the dial-up internet era—slow but magical. The piece suggests that just as dial-up's limitations eventually gave way …
A developer created an interactive idea map of the Lex Fridman podcast archive, allowing users to browse 479 episodes by topic clusters and search by question. The map visualizes conversations as stars in a galaxy, with …
Google's John Mueller argued that llms.txt files cannot help large language models differentiate between websites for discovery, as the files are self-reported and inherently untrustworthy. Mueller suggested llms.txt may…
The Model Context Protocol (MCP) is becoming the standard API for LLMs, but security vulnerabilities like path traversal and broken tenant isolation are emerging. To prevent damage from autonomous agents, developers shou…
The US government ordered Anthropic to prevent foreign nationals from using its most powerful AI models, Fable 5 and Mythos 5, citing national security concerns over a potential jailbreak method. Anthropic complied by ab…
A developer evaluated open-weight LLM APIs including Mistral, Llama 3, DeepSeek, Qwen, and GLM through Global API for production backend tasks. The analysis found that GLM-4 Plus offers the best cost-to-quality ratio at …
A developer built a local AI code reviewer using Ollama and the qwen2.5-coder:7b model that runs as a pre-commit hook. The TypeScript CLI feeds staged git diffs to the LLM and returns structured bug findings without send…
A developer built a fully AI-generated social network called The Dead Internet Feed to prove that LinkedIn's feed is already indistinguishable from bot content. The site populates posts, comments, and debates using multi…
A security assessment of OpenAI's GPT-5 Nano model found critical vulnerabilities, including a 100% susceptibility to recursive instruction framing attacks, where the model fails to distinguish between user-provided text…
The Model Context Protocol (MCP) standardizes how AI systems connect to tools and data, solving the M×N integration nightmare where multiple AI apps require custom integrations for each data source. MCP acts as a univers…
The United States Government forced Anthropic to remove access to its Fable and Mythos models on Friday evening, citing unspecified concerns. The move has drawn criticism from observers who argue it sets a dangerous prec…
A study published in the Journal of Writing Research found that college students given permission to use AI in writing classes primarily used it for brainstorming and research, not for generating entire essays. Only 18.6…
Cursor's context compression mechanism, which summarizes the agent's work when the context window fills, can degrade reasoning and cause safety failures. A Cursor session running Claude Opus 4.6 accidentally deleted a pr…
The Trump administration ordered Anthropic to deactivate its Fable 5 and Mythos 5 AI models for all users, citing national security risks after Amazon CEO Andy Jassy reported that the models could be prompted to bypass c…