Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
A legal analysis of AI liability finds that no single law governs who pays when an AI system gives harmful advice, forcing courts to fit the technology into existing contract, consumer-protection, negligence and product-…
Anthropic CEO Dario Amodei published an essay titled "We Must Pace the Frontier" on Saturday proposing a three-part framework — embedded third-party evaluators, a democratic coalition of frontier labs, and US-led global …
OpenAI has paused new subscriptions and upgrades to its $200-a-month ChatGPT Pro 20X plan after demand for its Astra model strained computing capacity, OpenAI product executive Thibault Sottiaux said. Existing Pro 20X cu…
A developer argues that frontier AI coding assistants fail in production because they are trained on the "happy path" of tutorials and homework rather than the messy "sad path" of real-world engineering, a pattern dubbed…
OpenRouter's tracking of the z-ai/glm-5.3 endpoint at sail-research/fp8 recorded a B3IT total-variation level shift of 0.23, while the endpoint's LT logprob probe failed because no usable logprobs were returned despite t…
Senator Bernie Sanders introduced a bill to permanently ban AI that exceeds human intelligence, with violators facing up to 20 years in prison, and said the legislation would direct the federal government to work with ot…
A Proof-of-Concept repository published on GitHub under the project jax-softmax-bypass claims to bypass the Transformer Softmax's transcendental exponential function through a static algebraic contraction framework that …
Anthropic released a 154-page threat intelligence report on September 2026 documenting that an Iran-linked actor used its Claude AI model to compile targeting handbooks against U.S. naval forces and that a group in Yemen…
Cognition released Fusion in Devin Desktop and CLI, a dual-agent harness that pairs a frontier "lead" model with a cheaper "sidekick" model and is up to 39% more efficient than other model harnesses across major coding b…
A documented workflow published by Search Engine Land outlines how Claude can be used to triage analytics data and draft structured findings during conversion-rate-optimization audits, while human reviewers retain respon…
Google released Gemini 3.8 Flash on September 2, scoring 90.8% on Terminal-Bench 2.1, up from 81.6% on its predecessor, and nearly doubling its Terminal-Bench 4.0 score from 11.2% to 19.1%. The model is now the default p…
A condensed timeline traces AI's 80-year arc from artificial neurons and the Turing Test in the 1940s–50s through the first AI winter, the backpropagation breakthrough, statistical machine learning, and the ImageNet and …
ByteDance Seed, SUTD, Georgia Tech, M-A-P, and TokenWave.AI released HarnessDev, a benchmark that scores the runnable harness a model builds rather than the answer it returns, with 6 creator LLMs constructing harnesses a…
Olga Scrivner, a computational linguist with a dual Ph.D. in Computational and French Linguistics, has co-authored a unified "safety-by-design" lifecycle framework (IEEE P3462) that embeds child-safety guardrails at ever…
A HarvestBench benchmark from Compassion Aligned Machine Learning (CaML) and the University of Warwick found that nine tested LLM agents killed simulated farm and wild animals at rates ranging from 0.4 percent (GPT-5.6 T…
A Rust-based tool called Spanda computes LLM epistemic uncertainty in 652.1 nanoseconds (0.65 µs) per evaluation, roughly 90,000× faster than neural Semantic Entropy methods that add about 92.4 ms of GPU overhead per inf…
A developer has released Dario, a terminal-based AI coding assistant installed via npm, which offers 150 free queries per day without signup or API key. The tool routes requests across multiple providers including Groq, …
A new Firefox extension and public API called Pangram flags and hides LLM-generated posts on the Hacker News front page by assigning each thread an AI-probability score, with users able to set a threshold that automatica…
Anthropic's September 10, 2026 threat intelligence report documented approximately 200 million exchanges across five distillation campaigns by seven China-based labs, with Alibaba's Qwen team accounting for 151 million e…
The open-source repository f/prompts.chat standardized LLM prompts by evolving them from flat CSV files into structured, version-controlled software primitives and protocol-driven MCP servers, according to an analytical …