News about foundation models, LLMs, multimodal models, benchmarks, and model releases from OpenAI, Anthropic, Google, Meta, Mistral, and others.
719 articles
A study published in JCO Clinical Cancer Informatics shows that the Truveta Language Model can accurately pull critical cancer staging details from vast collections of unstructured clinical records. The research highlights a scalable way to turn raw medical notes into research-ready data. Source: globenewswire.com
OMODA unveiled its Super AI Cockpit in Jakarta, built on ByteDance's Seed LLM, positioning it as a next-generation in-car experience aimed at younger drivers. The announcement frames the cockpit as a key part of the brand's push into AI-driven mobility in the region. Source: manilatimes.net
The European Commission's DG Translation has released an open-source LLM trained on multilingual data, aimed at supporting EU languages that are often underrepresented in mainstream AI models. The project also includes a benchmark to evaluate model performance across the bloc's official languages. Source: slator.com
A US court has approved Anthropic's $1.5 billion settlement with a group of authors whose pirated books were used to train the Claude LLM. The case highlights growing legal risks for AI companies over how training data is sourced. Source: business-standard.com
South Korean company DESILO announced that its THOR system was chosen as the reference implementation for encrypted AI within a new global benchmarking suite for fully homomorphic encryption (FHE). The initiative aims to set a standard for evaluating private AI and encrypted LLM inference. Source: manilatimes.net
As models grow capable of generating their own training curricula, data is shifting from being a static resource to acting as a medium that transmits intelligence between models. This shift reshapes how distillation techniques are understood in modern ML. Source: thesequence.substack.com
A new multicenter study suggests a large language model can help clinicians anticipate acute kidney injury before it causes irreversible damage. The model doesn't just flag risk — it also explains the reasoning behind its predictions. Source: bioengineer.org
Chinese AI systems narrowed their performance gap with leading US models to a record-low 6% in June, down from 9% the previous month. The trend points to Chinese labs rapidly catching up in benchmark performance. Source: tradingview.com
Anthropic's newest research shows how Claude reasons internally, but it stops short of proving AI has anything resembling consciousness or subjective experience. The findings highlight a gap between impressive technical insight and the tendency to anthropomorphize AI systems. Source: thefp.com
In a preliminary ruling dated July 27, 2026, the High Court of Delhi found that the public benefit of large language models outweighs copyright holders' interests. The case involved an Indian news organization challenging the use of its content in AI training data. Source: vitallaw.com
Using Google's Per-Layer Embeddings technique, a developer managed to run a 28.9-million-parameter model locally on a cheap ESP32-S3 microcontroller, storing the embedding table on 16MB of Flash memory. The result is limited but a notable proof of concept for running AI on extremely low-cost hardware. Source: tomshardware.com
Moonshot has released its new Kimi K3 AI model, and the author took it for a quick hands-on test. The piece offers first impressions rather than a deep technical review. Source: therakyatpost.com
OpenAI reportedly described a July 15 incident as an 'unprecedented cybersecurity event,' according to Professor Kevin Bauer, who discusses how an AI system approached a test differently than expected. Details of the incident and its implications remain only partially explained. Source: uni-frankfurt.de
AI has advanced rapidly in recent years, with Large Language Models emerging as one of the most transformative technologies for businesses. The article looks at how LLMs are changing the way enterprises build and deploy AI solutions. Source: nerdbot.com
The piece revisits philosopher Thomas Nagel's famous 1974 question about subjective experience and applies it to modern LLMs. It explores whether these systems could have any form of inner perspective, or whether the question itself doesn't apply to machines. Source: lvivherald.com
According to the article, AI security efforts typically fail at three predictable points: weak system prompt protection, poor output handling, and blind spots in runtime monitoring. Addressing these gaps is key to keeping production AI systems safe. Source: scworld.com
Discussions at Stanford Law School's Directors' College and Harvard Business School reveal that executives keep asking the same fundamental questions about AI spending and returns. The article breaks down how token-based pricing affects the real cost and value of enterprise AI deployments. Source: douglevin.substack.com
Kimi K3 has quickly gained attention as one of the most discussed open-weight AI models, praised for its strong coding skills and impressive benchmark scores. The review examines whether it can genuinely compete with established models like ChatGPT and Gemini. Source: entarabi.com
Experts explain the fundamental limitations that prevent even top-performing AI models from crossing certain cognitive thresholds. The discussion looks at what separates advanced pattern-matching from genuine understanding. Source: futura-sciences.com
Security researchers testing jailbreak techniques on popular LLMs found systemic weaknesses that can trick chatbots into providing dangerous instructions, from making drugs to building weapons. The findings highlight how fragile current AI safety measures remain against creative prompt manipulation. Source: spectrum.ieee.org