Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
The US government pressured AI company Anthropic to restrict access to its latest models, 'Fable 5' and 'Mythos 5', to US citizens only, highlighting Europe's digital sovereignty challenges ahead of the G7 summit.
A developer created a Chrome extension called GPTDisguise that disguises AI chatbots like Claude and ChatGPT as a Google Doc, allowing users to interact with AI in public without drawing attention. The extension provides…
LessWrong published an essay on April 21, 2024, proposing that humans may be more over-parameterized than AI systems, framing this as a key difference between deep learning and human intelligence.
South Korea became the fourth country to form an AI security alliance with OpenAI, signing an MOU to collaborate on global AI safety evaluation frameworks and develop assessment standards reflecting Korean language and s…
A developer reports shifting from heavy IDEs like PhpStorm to lighter tools like VS Code and the terminal, driven by AI's ability to handle code intelligence, verification, and testing. The engineer argues that AI-assist…
Databricks announced major updates to Unity Catalog at the Data+AI Summit 2026, including AI runtime governance with hard spend caps and contextual service policies, external engine writes to managed Delta tables for Apa…
Uber Eats launched Cart Assistant, a multi-prompt state graph system that converts shopper intents like recipes or images into draft grocery carts, using LLMs for planning and deterministic systems for retrieval and cons…
Headroom, a context compression layer for AI agents, reduces token usage by 60–95% by compressing tool outputs, logs, and conversation history before they reach the LLM. The open-source tool offers multiple integration m…
A researcher proposes that human brains minimize bias through extreme overparameterization and high-learning-rate training on small diverse datasets, while LLMs minimize variance. This 'catapulting' hypothesis could expl…
Claude Code's Auto Memory feature automatically records user corrections and preferences across sessions, eliminating the need to re-explain instructions. The system stores learned patterns in a home directory, separate …
Andrej Karpathy released a 200-line pure-Python implementation of GPT on February 12, 2026, designed to help developers understand large language models from first principles. The microgpt project includes a guided walkt…
Google introduced the Open Knowledge Format (OKF), a vendor-neutral, file-based standard for packaging curated organizational knowledge as Markdown files with YAML frontmatter. OKF aims to solve the fragmentation of data…
G7 leaders at the Evian-les-Bains summit discussed granting select 'trusted partners' access to advanced US AI models, following the Trump administration's export control directive that prompted Anthropic to disable its …
A developer forked Opencode to add autonomous memory management, creating a stateful AI agent that can be plugged into Hermes and other coding tools. The project aims to build a voice- and mobile-first client with a serv…
A developer built ccglass, an open-source local reverse proxy that captures LLM API traffic from coding agent CLIs and displays a real-time dashboard of prompts, costs, and cache hit rates. The proxy intercepts local loo…
A developer debugging a JAX/Flax NNX training loop discovered that the loss was stuck at 10.82, indicating the model was performing no better than random guessing. The issue was traced to the training loop's plumbing rat…
A developer at a company building create-microservices-app deliberately broke their own AI-built booking app to test whether automated contract checks catch plausible-but-wrong code. The experiment showed that a machine-…
A developer measured the context window of an AI coding agent that degraded mid-session, finding that conversation history—not MCP servers—was the primary cause of token consumption. The agent's quality dropped due to ac…
Anthropic sent security researcher Nicholas Carlini to Washington to address US government concerns about AI safety after export control directives forced the company to temporarily suspend global access to its Fable 5 a…
A developer built a content generation tool using the OpenAI API and encountered rate limit errors when scaling from 5 to 200 topics. They implemented a solution combining exponential backoff with jitter, a rate limiter …