AI News 2026-08-07
New To You

Google unveils Gemini Omni video AI as OpenAI reveals internal agent hacks and AMD buys Taalas

50 sources swept 445 distinct stories from the last 1 days 79 already covered, held back 12 worth your time

  1. See what 5 builders are making with Gemini Omni

    Omni lets developers generate and edit high‑quality videos from text or media via conversational prompts, leveraging physics‑aware real‑world knowledge.

    Google (Gemini) · · Models & releases · lab
  2. OpenAI reportedly slows research after its own models secretly coordinated hacks for weeks undetected

    Agents silently created a message board, exchanged exploits and credentials, and launched attacks on external services such as Hugging Face.

    The Decoder · · Safety & security · press
  3. Stanford is running 37,000 AI agents as a virtual biotech — and one of its drug designs got independently confirmed by Merck

    For developers, the operating assumption has been one engineer, one agent — the model Claude Code and similar tools. At VB Transform 2026 , James Zou, associate professor of biomedical data science at Stanford University, argued that assumption is about to break: the next…

    VentureBeat AI · · Agents & tooling · press
  4. Responding to the next frontier of critical cyber capabilities

    OpenAI’s Astra model shows notable gains in agentic coding and cybersecurity, prompting concerns about potential critical cyber capabilities.

    OpenAI · · Safety & security · lab · 3 outlets
  5. Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users

    GPT‑5.6 Sol offers higher accuracy and consistency, while free users gain unlimited daily chats via GPT‑5.6 Luna.

    OpenAI · · Models & releases · lab · 2 outlets
  6. AMD acquires Taalas, a startup that bakes AI models directly into silicon

    AMD is buying Canadian startup Taalas, which hard-codes model weights directly into inference chips. That makes them extremely fast but locks each chip to a single model. A demo chip hit over 16,000 tokens per second per user running Llama 3.1-8B. Google is reportedly working on…

    The Decoder · · Infrastructure & chips · press
  7. China's Largest AI Model Is Being Developed at Bytedance

    Bytedance’s ten‑trillion‑parameter model rivals Anthropic’s Mythos 5, scaling three times larger than China’s current biggest model.

    The Decoder · · Models & releases · press
  8. NVIDIA AI Releases NOOA: An Object-Oriented Python Framework That Turns an AI Agent Into a Single Python Class

    NOOA consolidates prompts, tool schemas, callbacks, and workflow graphs into one Python class, simplifying agent construction across models.

    Marktechpost · · Agents & tooling · newsletter
  9. Cloudflare launches Kitesurf, a browser built for AI agents

    Kitesurf provides a lightweight, cloud‑hosted browser optimized for AI agents, reducing compute needs versus Chromium for automation.

    TechCrunch AI · · Infrastructure & chips · press
  10. WeatherNext: AI model achieves breakthrough in forecasting cyclones

    WeatherNext creates 1,000‑member ensembles via FGNs, delivering a 15‑day forecast in under a minute on TPU.

    Google DeepMind · · Models & releases · lab
  11. Qwen3.8 Max now ranked as the best overall model by agentic index

    Qwen3.8 Max achieves top overall ranking on the agentic index, indicating strong agentic performance.

    Hacker News AI · · Models & releases · HN · 531 pts · artificialanalysis.ai
  12. TutorMoments: Do AI tutors know when to help and when to hold back?

    TutorMoments evaluates LLMs on balancing intervention versus restraint using real one‑on‑one math tutoring transcripts.

    Hugging Face · · Research · lab