The news desk
Recent developments in artificial intelligence, with the original sources one click away.
Researchers publish evaluation frameworks for medical LLMs in ophthalmology, personalized AI feedback, and mental health chatbot safety assessment.
Sources · arXiv AI (cs.AI)
A16z dedicates $1.7 billion of its new $15 billion fund to its infrastructure team handling major AI investments.
Sources · TechCrunch AI
Anthropic has launched Claude Opus 4.6, its latest flagship AI model, boasting significant advancements in financial analysis, enterprise workflows, and coding.
Sources · Bloomberg Law / The New Stack +5
OpenAI shares case study of family using ChatGPT alongside medical professionals to navigate their son's cancer treatment options.
Sources · OpenAI
French AI startup Mistral launches new translation model, claims efficiency advantage over larger US competitors with fewer GPUs.
Sources · WIRED AI
Recent arXiv papers explore distilling multi-agent systems, conversational safety risks, and reinforcement learning assumptions in LLM training.
Sources · arXiv AI (cs.AI)
Researchers publish papers on improving inference-time computation for language models, diffusion models, and audio-language emotion recognition.
Sources · arXiv AI (cs.AI)
BGL, serving 12,700+ businesses across 15 countries, explores AI-driven business intelligence using Claude Agent SDK and Amazon Bedrock AgentCore.
Sources · Amazon AWS AI
Three new arXiv papers address LLM mathematical reasoning, inference optimization, and training stability challenges.
Sources · arXiv AI (cs.AI)
Recent preprints introduce methods for maintaining LLM safety during fine-tuning, calibrating AI trust, and teaching LLMs to forecast time series data.
Sources · arXiv AI (cs.AI)
New arXiv papers explore privacy in compressed LLMs, collective false memories in AI agents, and retrieval-augmented generation for medical diagnostics.
Sources · arXiv AI (cs.AI)
Apple's Xcode 26.3 adds agentic coding features through integrations with Anthropic's Claude Agent and OpenAI's Codex.
Sources · TechCrunch AI
Google adds Poker and Werewolf to its Game Arena AI benchmarking platform, with Gemini 3 Pro and Flash leading the chess leaderboard.
Sources · Google AI Blog
The Department of Health and Human Services has employed AI tools from Palantir and Credal AI to identify DEI and gender ideology language in grants.
Sources · WIRED AI
Mistral AI advocates for collaborative design of tailored AI solutions with enterprise partners after many early generative AI pilots failed to deliver value.
Sources · MIT Technology Review
Three new arXiv papers tackle adaptive reasoning strategies, long-context inference optimization, and reinforcement learning training instabilities.
Sources · arXiv AI (cs.AI)
Recent research explores privacy in compressed LLMs, collective false memories in multi-agent systems, and diagnostic reasoning with retrieval-augmented generation.
Sources · arXiv AI (cs.AI)
Indonesia has lifted its ban on xAI's Grok chatbot, joining Malaysia and the Philippines in reversing restrictions on the AI tool.
Sources · TechCrunch AI
Microsoft internally forecasts water consumption at its data centers will more than double by 2030 compared to 2020 levels, driven by AI growth.
Sources · NYT Technology
Recent arXiv papers examine transformer arithmetic learning, reinforcement learning for LLM reasoning, and vision language models for detecting AI-generated images.
Sources · arXiv AI (cs.AI)