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 Trump administration has asked OpenAI to delay the release of its GPT-5.6 model, citing security concerns. OpenAI CEO Sam Altman told employees the company will release the model in limited preview with government-ap…
A developer built CORE, a cognitive system using Cl(4,1) Conformal Geometric Algebra to make incoherence algebraically impossible. Unlike large language models that monitor and correct for coherence, CORE enforces a vers…
Puter.js is a serverless JavaScript library that enables developers to build full-stack, AI-powered applications directly in the browser without needing a backend to protect API keys. It allows frontend apps to use OpenA…
A researcher is developing fab, an interface to help human researchers make sense of research produced by many AI agents running in parallel, focusing on automated alignment research. The project aims to augment human ju…
GitHub's Copilot code review now uses CLI-based file exploration tools, reducing costs by 20% while maintaining quality. Medium analysis depth public preview adds organization-level default settings and attribution label…
A study from the University of Edinburgh and UCLA found that large-language-model trading strategies underperformed a simple buy-and-hold approach over 20 years across 100+ stock symbols. The AI bots were too cautious in…
The U.S. government imposed export controls on Anthropic's Fable 5 and Mythos AI models, the first such controls on an AI model, citing a narrow jailbreak vulnerability also present in competing models. The ban forces An…
A developer reports that using a top-tier AI model to write a detailed specification and a cheaper model to implement from that spec yields better accuracy and lower cost than using the strong model alone without a struc…
Hong Kong's bilingualism is uniquely indispensable in the AI era, as it allows users to cross-reference information across English and Chinese data sets to detect AI hallucinations and epistemic asymmetries. A user demon…
Votee AI, led by CEO Pak-Sun Ting, is building sovereign, enterprise-grade AI architecture for underrepresented languages, starting with a native Cantonese model. The company aims to empower billions of underserved globa…
A developer created a memory cleaner for Claude that prunes bloated memory files through an interview-like diff process, requiring user approval for each change. The tool addresses performance degradation and contradicti…
OpenAI released new economic research showing a dramatic shift toward AI agents, with internal data from its Codex app revealing that 85% of employee output tokens now come from agents rather than chatbots. The study fou…
The Linux Foundation announced on June 3, 2026, its intent to launch the Tokenomics Foundation, dedicated to open standards for AI cost management, with support from Google, Microsoft, Oracle, JPMorganChase, and others. …
The New York Times has amended its copyright infringement lawsuit against OpenAI and Microsoft, dropping one claim against OpenAI and modifying another against Microsoft. The case, which alleges unauthorized use of milli…
Microsoft introduced new Copilot features in Excel for finance professionals, including skills for repeatable workflows, financial connectors for trusted data, and improved traceability. The tools were tested by Microsof…
Autonomous AI agents are moving from experimental playgrounds into enterprise application architecture, requiring robust design patterns and security measures. A developer with 20+ years of experience outlines core compo…
OpenAI CEO Sam Altman told staff on June 24 that GPT 5.6 will first be released as a limited preview to a small group of partners because the federal government asked OpenAI to stagger the rollout, according to The Infor…
OpenAI has agreed to a government-requested slow rollout of its GPT-5.6 model, delaying public release. The decision does not affect development pace but widens the gap between internal and public access. Prediction mark…
OpenAI published a paper claiming 98% of its employees now use its Codex AI coding agent, up from 40% in August 2025, and that non-developer usage grew 137 times. However, all metrics are self-reported by OpenAI, which s…
Inferize announced DeepSeek-V4-Pro, claiming it can serve the model in 20 seconds with highly optimized, elastic AI inference. The company is building fast, efficient LLM serving that scales with demand and has opened a …