The news desk
Recent developments in artificial intelligence, with the original sources one click away.
Amazon and researchers explore scalable methods for fine-tuning large language models using integrated platforms and evolution strategies.
Sources · Amazon AWS AI / arXiv AI (cs.AI)
OpenAI will not use 'io' name for its AI hardware, expected to ship in 2027, according to court filing in trademark dispute.
Sources · WIRED AI
ArXiv publishes research on LLMs for finite-state reasoning, diffusion models for search agents, and token-level model collaboration.
Sources · arXiv AI (cs.AI)
Recent research explores LLM safety degradation during fine-tuning, secure code generation methods, and the mathematical basis of parameter-efficient training.
Sources · arXiv AI (cs.AI)
Svedka debuts first AI-generated Super Bowl ad while Anthropic references OpenAI rivalry in Big Game commercials.
Sources · TechCrunch AI
Recent arXiv papers tackle training data selection, reasoning distillation benchmarks, and reward modeling for specialized LLM applications.
Sources · arXiv AI (cs.AI)
Recent arXiv preprints explore edge-scale LLM agents, social deduction games for multi-agent evaluation, and selective fine-tuning methods.
Sources · arXiv AI (cs.AI)
Recent arXiv papers examine French language model performance, LLM safety under fine-tuning, and AI chatbot safety for mental health support.
Sources · arXiv AI (cs.AI)
Tech entrepreneurs in San Francisco pursue AI business opportunities while questioning the technology's broader societal implications.
Sources · NYT Technology
Large language models struggle with research-level math problems, requiring human mathematicians to evaluate their poor performance.
Sources · NYT Technology
Moltbook, a social network designed for AI agents, suffered a data exposure incident affecting real human users' information.
Sources · WIRED AI
Anthropic is relying on its AI model Claude to develop the understanding needed to prevent potential AI-related disasters as systems advance.
Sources · WIRED AI
Moltbook, a Reddit-like platform where AI agents interacted while humans observed, briefly captured internet attention this week.
Sources · MIT Technology Review
Three new research papers evaluate AI language models for medical applications, focusing on ophthalmology, personalization, and mental health safety.
Sources · arXiv AI (cs.AI)
OpenAI outlines its localization approach for making frontier AI models work across different languages, legal frameworks, and cultural contexts.
Sources · OpenAI
Researchers propose ProAct for long-horizon planning, RGCF for explainable recommendations, and LittleBit for sub-1-bit quantization of large language models.
Sources · arXiv AI (cs.AI)
Google adds AI-powered auto-browse to Chrome, while open source Moltbot attracts users and Qwen3-Max-Thinking makes its debut.
Sources · Last Week in AI
OpenAI's GPT-5 integrated with Ginkgo Bioworks' automation achieved 40% cost reduction in protein synthesis through autonomous experiments.
Sources · OpenAI
Entertainment featuring or created with AI has failed to connect with audiences over the past year, according to industry observations.
Sources · WIRED AI
MIT Technology Review explores a commonly misinterpreted visualization related to frontier large language model releases from major AI labs.
Sources · MIT Technology Review