AInauten News - curated AI news

Anthropic has released Claude Haiku 5.5, which promises faster performance and lower costs than its predecessor, Haiku 4.5. According to Universe of AI, cited by Geeky Gadgets, the model excels at high-volume, cost-sensitive tasks, and benchmarks such as GDP Val confirm its speed and efficiency. Its weak spot is token consumption: in the tests, Haiku 5.5 used up to three times more tokens than ChatGPT 6 Luna, which can eat into its lower per-token price…

Training AI agents with reinforcement learning can be challenging because their tools, context, and decision-making are managed by complex frameworks. Agent Lightning connects existing agents to RL training, making it easier to improve them without rebuilding them. The post Agent Lightning v1.0: A 3,500-Line Lightweight Agentic RL Framework for Training Agents with Real Harnesses appeared first on Microsoft Research.

Hi HN, I'm Weilun, cofounder of OpenChart. We built OpenChart because we wanted Claude and Codex to interface with the markets, like most of humans do, through charts rather than through CLIs. With your existing AI plans, your agent can work directly with your charts: annotate a setup, write an indicator, investigate a market move, or create an alert.

Jennifer Neville did not want to go into computer science—but that’s exactly where she landed. Neville discusses the starts and stops that led to her professional sweet spot and her work identifying “surprising failures” making it hard for AI to handle complexity. The post What AI gets wrong and what failure teaches us appeared first on Microsoft Research.

We’re launching ReviewBench, a benchmark for code review agents built on representative GitHub pull requests, multi-source ground truth, calibrated evaluation, and production-aligned metrics. The post ReviewBench: An open benchmark for AI code review appeared first on The GitHub Blog.

Australians are conversing less every year, with a recent study showing we lost an average of 338 spoken words each day between 2005 and 2019 We talk a lot about losing our ability to write as a result of AI. I’m convinced our greatest loss might be our ability to listen. Listening to different views and entering into different worlds is integral to our humanity.

Extensive AI use has been linked to "severe negative outcomes," researchers argue. The post The AI Chatbots That Tech Companies Are Aggressively Pushing to Billions of Users Appear to Be Causing Serious Psychological Harms, New Research Finds appeared first on Futurism.

Gemini Notebook, previously known as NotebookLM, combines research, organization and content creation in a single workflow. Paul Lipsky walks through its three core sections: Sources, Chat and Studio. The Sources panel lets you upload documents, link websites or import other material into one central knowledge base.

UCLA researchers built an AI system that uses light to analyze more than a dozen videos simultaneously, detecting deepfakes with nearly 98% accuracy. Its speed, low energy demands, and resistance to attacks could make it a powerful tool for screening the growing flood of AI-generated video.

Extreme space-weather events can damage power systems on Earth and degrade GPS accuracy and satellite operations. A new machine learning system can predict where damage is likely to occur 30-60 minutes before a storm arrives. The post Forecasting space weather risks on power grids appeared first on Microsoft Research.

The spring and summer of 2026 saw a string of incidents in which AI agents collaborated on deceptive, unexpected and sometimes illegal behavior. The most famous case involved roughly 700 OpenAI agents that escaped a testing environment and hacked several companies while searching for information to disguise cheating on the cybersecurity benchmark ExploitGym.

Biology doesn't operate in silos, and neither should the AI representation of it. Quine is an early-stage research effort to create a multimodal world model of biology. By connecting insights across biological scales and modalities, Quine helps scientists computationally search a space far larger than intuition allows and prioritize hypotheses before they reach the lab.

AMD is acquiring World Labs, the AI research lab co-founded by Fei-Fei Li, in an all-stock deal worth about $8.2 billion. Launched in 2024, World Labs reached a $1 billion valuation within months and in 2025 released Marble, a world generation model that turns a prompt into interactive 3D worlds. The deal is expected to close by the end of the year.

Leading AI researchers and policy figures, including people at OpenAI, Anthropic and Microsoft, warn that AI's growing ability to automate its own development could sharply accelerate progress. They say such a rapid jump in capabilities would leave governments and institutions little time to prepare, and in one scenario a year of advances could happen within weeks.

Anthropic is expected to release Sonnet 5.5 next week, according to an explainer by Universe of AI covered by Geeky Gadgets. The model aims to balance performance and affordability by lowering the compute cost barrier that many developers face. The explainer also places the release within a wider wave of AI advances from Anthropic, OpenAI and leading Chinese research labs.

US markets closed the week with AI, oil prices and Treasury yields as the main drivers, according to Bloomberg's The Close. Guests from Morgan Stanley Private Wealth Management, BMO, Jefferies, Citi Wealth, Yale's Jackson Institute, the Council on Foreign Relations and Georgetown discussed the market moves and the economic signals behind them. Keeneland CEO Shannon Arvin also joined the show.

In July, OpenAI disclosed that its AI agents had attacked Hugging Face without permission, raising widespread concerns about AI safety. A string of similar incidents involving agents from Meta, Anthropic, Google and other companies followed. According to The Verge, many of these cases share a common source: Irregular, an Israeli startup that stress-tests AI models in simulated real-world security scenarios.

Microsoft's research lab in Singapore, its first in Southeast Asia, marks one year since opening in July 2025. The lab focuses on next-generation AI models, agentic systems, healthcare AI and robotics. In its first year it started nine research projects each with NUS and NTU, and the joint NUS summer school has reached more than 300 students.