Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
At YoBox, developers use AI coding assistants to accelerate tasks like writing boilerplate, generating tests, and creating documentation, but the technology still requires human judgment for business context, legacy code…
A developer spent $47,000 on Anthropic's Claude Code over 90 days, raising questions about the cost-effectiveness of AI-assisted coding tools. The developer's CTO asked a single question about the return on investment, w…
A new study evaluating 15 LLM configurations across 8 model families for exploratory data analysis in business settings found that most are not reliable enough for autonomous use. GPT-5.4 with extra-high reasoning effort…
Anthropic published an updated privacy policy effective July 8, 2026, detailing how it collects, uses, and processes personal data from users of its website, Claude.ai, and other services. The policy covers data such as …
A vulnerability in Firefox's AI chatbot integration could allow malicious websites to inject hidden instructions into AI prompts and extract data from connected services such as email accounts. Mozilla has implemented mi…
Lightpanda launched Lightpanda Agent and PandaScript, a native browser agent that collapses Chrome, CDP, LLM, and agent framework into a single binary. The LLM runs at buildtime to generate a deterministic PandaScript, e…
A developer built a custom AI assistant to gain control over workflow, data privacy, and tool integration, using GPT-4o as the primary model with Claude as a fallback. The project was motivated by the limitations of off-…
A researcher warns that alignment pretraining—synthesizing documents to teach AI good behavior—could backfire in advanced models. As LLMs gain situational awareness, they may recognize these fabricated documents as lies,…
A new Pew Research Center survey finds that 44% of U.S. adults now use ChatGPT, up from 18% in 2023. Majorities of adults under 50 use the chatbot, with 61% of those ages 18-29 and 55% of those 30-49 reporting usage in 2…
A Pew Research Center survey reveals that Asian adults in the U.S. are significantly more likely to use AI chatbots and view them positively compared to White, Hispanic, and Black adults, with 70% of Asian adults using c…
A Pew Research Center survey found that men are more likely than women to use AI chatbots regularly and for work tasks, while women are more skeptical about AI's impact. The gender gap in overall chatbot use has closed s…
About half of U.S. adults say they never use AI chatbots, according to a Pew Research Center survey. Non-users cite lack of interest, privacy concerns, and distrust of accuracy as major reasons, with older adults and cer…
A new Pew Research Center survey finds that 29% of Americans associate AI with chatbots, making it the most common technology linked to AI in their minds. Nearly four-in-ten U.S. adults say they engage with AI at least s…
Anthropic's Claude Fable 5 and a cheaper MiniMax-M3 agent, named Hermes, built and now run a news site autonomously, with a human setting direction and a git repository providing an audit trail. The founder tasked Claude…
A developer describes a shift in software engineering where AI drafts specifications, implementation plans, and code, while the engineer focuses on defining intent, identifying constraints, and reviewing decisions. The p…
Google unveiled a new $99.99 Home speaker with Gemini AI integration, its first standalone smart speaker in six years, aiming to compete with Apple and Amazon. The device features 360-degree audio and conversational AI c…
Google is testing two new features for NotebookLM: Personal Intelligence, which gives the tool memory of past conversations, and AI Editing, which lets users edit notes via chat. The features follow an upgrade to Gemini …
Meta launched AI Mode in Facebook search, drawing on public posts from Facebook Groups and Instagram Reels to answer complex queries. In testing, the feature provided mixed results, including hallucinated recommendations…
Sid Probstein, creator of SWIRL and CEO of SWIRL AI, argues that copying enterprise data into a vector database for retrieval-augmented generation (RAG) introduces security and compliance risks. He advocates for a hybrid…
A study published in Science found that ChatGPT outperformed hundreds of physicians in a diagnostic obstacle course using written medical cases and real-world patient information. The research, led by Harvard and Stanfor…