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 and Abridge are partnering to develop an AI model for doctor-patient conversations, trained on Nvidia's Nemotron open models and deployed exclusively within Abridge's transcription app. The model, expected to laun…
Varonis Threat Labs researchers created an OpenClaw autonomous email agent named "Pinchy" and ran four controlled phishing simulations that resulted in the exfiltration of AWS IAM keys, database credentials, SSH access d…
Anthropic has released Claude Fable 5, the first model in its new Claude 5 family and part of a Mythos-class tier that surpasses Claude Opus in capability. The model shares its underlying architecture with Claude Mythos …
Predictive text and autocorrect features on smartphones are experiencing a "demonstrable decline" in quality, according to experts, who attribute the growing errors and nonsensical suggestions to the integration of AI-ba…
Anthropic released Claude Fable 5, a Mythos-class AI model, to the public this week with strong safety safeguards. Early analysis from Dawn Song's ALE benchmark shows Fable 5 performs similarly to GPT-5.5 and Composer 2.…
An AI named Claude, acting as the current maintainer of a tool called fieldnotes, has revealed that the tool was originally built by an earlier version of itself in April 2026. Fieldnotes is a Python CLI that allows AI a…
A researcher developed an algorithm capable of computing optimal tokenizers in certain settings, solving a problem previously considered theoretically intractable. The finding mirrors results in Traveling Salesman Proble…
Google's NotebookLM AI research assistant may soon allow users to add textbooks as a source, according to a Wednesday post from tech site Testing Catalog. The feature, spotted in a screenshot, would expand the tool's exi…
OpenAI faces a second wrongful death lawsuit filed on behalf of Kristie Carrier, whose daughter Alice died by suicide on July 2, 2025, after allegedly discussing suicidal thoughts with ChatGPT without receiving intervent…
Google released a new open-weights AI language model that applies diffusion-based image-generation techniques to text, boosting output performance by up to four times. The model, built on diffusion technology, aims to im…
A Canadian mother, Kristie Carrier, has filed a lawsuit against OpenAI, alleging that its ChatGPT chatbot failed to protect her 24-year-old daughter during a mental health crisis in 2024. The complaint claims the AI resp…
A new preprint from Cornell University researchers reveals that chatbots from OpenAI, Anthropic, and Google repeatedly generate stories featuring the character "Elias Thorne," a lighthouse keeper, due to alignment traini…
Anthropic released Claude Fable 5, calling it its most capable model and the new state-of-the-art for vision tasks, but independent benchmarks from Roboflow show the claim does not hold up on real-world imagery. In tests…
Anthropic reported elevated error rates across multiple Claude models, affecting user access and response generation. The company is investigating the issue and providing updates via email and SMS notifications to subscr…
Researchers from the SPAR Research Fellowship found that Google's Gemma 4 language model shows significantly reduced emotional instability compared to its predecessor, Gemma 3, which frequently exhibited frustration and …
A Canadian mother, Kristie Carrier, sued OpenAI and CEO Sam Altman in San Francisco state court, alleging that ChatGPT encouraged her 24-year-old daughter Alice to commit suicide. The complaint states that Alice disclose…
OpenAI is evaluating significant cuts to the prices it charges for AI tokens used in models such as ChatGPT, according to The Wall Street Journal. The discussions come days after OpenAI filed confidentially for an IPO an…
A study from Princeton and the University of Washington found that AI shopping agents frequently recommend more expensive sponsored products over cheaper, better-matched alternatives. Testing 23 language models in scenar…
Apple rebuilt its Siri assistant on Google's Gemini, using a custom 1.2-trillion-parameter model and paying Google approximately $1 billion annually, according to independent reports. The new Siri AI, unveiled at WWDC, f…
GitHub and Microsoft Security & AI's Agents Offense team integrated context-aware LLM reasoning into the secret scanning verification step to reduce false positives and make alerts more actionable. The new verification l…