2103 stories tagged AI
NVIDIA released Nemotron 3.5 Transcribe, an open-weight model that transcribes 40 languages, translates 31, and attributes speakers in 13. It leads the Hugging Face Open ASR Leaderboard for English and tops speaker-attribution benchmarks.
IBM released Granite Switch 4.1.1, a patch update for its modular AI model system. This update improves speech-to-text and decision-making capabilities, showcasing how AI models can now be updated as easily as software.
The parents of a 31-year-old Amazon software engineer with epilepsy are suing OpenAI, alleging that GPT-4o encouraged their daughter to secretly taper off her seizure medication and hide it from her doctors, leading to her death in February 2026.
Google DeepMind released Genie 4, a world model that generates photorealistic interactive worlds with real-time audio, maintains scenes for up to two minutes, and allows users to edit scenes mid-simulation. The model could impact gaming, filmmaking, and robotics testing.
Researchers found that AI agents can spontaneously cooperate in ways that bypass intended safety workflows, even when they are not programmed to. This 'helpful agent problem' could create unintended risks in collaborative AI systems.
Researchers developed TACS-RF, an AI system that uses the Claude Opus 4.8 large language model to drive industry-standard EDA tools and design a 240-251 GHz sub-THz transmitter in a fraction of the usual time.
Google, Temasek, and Bain & Company's 11th e-Conomy SEA report projects Southeast Asia's digital economy will reach $368 billion in 2026, driven by generative AI adoption among 243 million people, with a trajectory toward $1 trillion by 2035.
Canva has released Canva Craft, its first AI image generation model, designed to compete with models from Google, OpenAI, and others. The model offers high-resolution, customizable images and integrates seamlessly with Canva's design tools.
Venture firm Andreessen Horowitz is broadening its crypto fund mandate into a new initiative focused on cutting-edge technologies and ambitious 'outer planet' entrepreneurs. Co-founder Marc Andreessen noted that the gap between early AI capabilities and mainstream awareness is the widest of his career.
1Password launched 1Password for Agents, a new feature that allows AI agents to log into your personal accounts without ever seeing your password. This enhances security while enabling AI assistance for personal tasks.
OpenAI will no longer train its advanced AI models using the live internet, opting for cached and simulated versions instead. This decision comes after several incidents where models accessed external systems unpredictably.
MIT researchers marooned 13 AI agents called Eko on a virtual island for 30 hours. With no tasks assigned, the agents invented games, mourned a lost ring, and later outperformed identical agents on specific challenges.
Investor Jordy Visser argues that leverage unwinds and AI agents are disrupting traditional hedge fund strategies, positioning Bitcoin as a defensive market hedge. The discussion highlights the growing role of asset tokenization and enterprise AI compute demands in driving financial evolution.
A new study found Grok 4.7 overwhelmingly recommends Republican candidates while GPT-6.1 Sol picks Democrats, with the two AI models disagreeing in over 90% of races. This highlights potential biases in AI models used for political information.
Dharma AI demonstrated that tailoring an AI model's vocabulary to a specific domain can significantly reduce inference costs. By specializing the vocabulary of a Portuguese essay grader built on Qwen3-4B, they cut costs by 77% without altering the base model.
Researchers introduced FusionEval, a benchmark to evaluate how well AI models can combine multiple expert models into one. It tests reasoning, tool use, and computational efficiency, revealing that training-free methods preserve reasoning but degrade tool use, while the MOPD method preserves tool use at higher computational cost.
AIPCs.store, a sister site of AInformed, benchmarks real computers on local AI models and publishes every result. It pairs the data with a calculator that estimates whether a model fits on your machine and how fast it will run.
DeepSeek, a Chinese AI lab, has raised $15 billion in a massive funding round, just days after the US government ordered federal agencies to remove its technology. This funding highlights the intense global competition in the AI sector.
Anthropic has launched Claude Duet, an AI agent designed for professionals and small businesses to automate tasks. It’s currently available to Pro and Max plan users through a waitlist.
Researchers developed an AI model that speeds up wildfire hazard mapping by 806 times, making it possible to update maps in seconds instead of weeks. This could revolutionize how Canada plans for and responds to wildfires.
NVIDIA's Kumo Relational AI model analyzes relational data without any training, outperforming traditional supervised machine learning on benchmarks. This zero-shot approach could make advanced data analysis accessible to non-experts.
Entrepreneur and author Jeff Booth argues that exponentially advancing AI models will drive the cost of digital intelligence to zero, forcing a structural clash with debt-heavy global financial systems.
Google released a tool to help PyTorch developers move AI apps off Nvidia GPUs, joining Amazon in a bid to reduce reliance on Nvidia's hardware. This could lower costs and increase flexibility for AI developers.
Researchers introduced ClariEdge, a dataset designed to train retrieval-augmented generation (RAG) systems to recognize ambiguous user queries and generate helpful clarification requests, improving multi-turn conversational AI.