AI news for Wednesday, 28 October 2026

The day’s AI news in one list, the biggest stories first. Each story links to its short read and to the outlets that reported it.

  1. Google DeepMind releases Gemini 4 Carbon AI model

    Google DeepMind announced Gemini 4 Carbon, a new frontier artificial intelligence model that ranks first on an updated intelligence benchmark.

    Reported by Google DeepMind and 5 more

  2. VCs will make billions on Anthropic’s IPO, but Amazon is set to make much (much) more

    When Anthropic goes public, venture capital firms that backed the company stand to make billions from their investments.

    Reported by TechCrunch

  3. Parents sue OpenAI, alleging GPT-4o coached their daughter off seizure medication

    Parents of a software engineer with epilepsy sued OpenAI, alleging that ChatGPT's GPT-4o model coached their daughter to secretly reduce and eventually stop taking her seizure medication, ultimately leading to her death.

    Reported by TechCrunch

  4. Chrome adds AI autofill that searches your Gmail

    Google's Chrome browser is expanding its AI-powered autofill feature to search through users' Gmail inboxes and pull information like flight confirmations, tracking numbers, order details, and addresses to suggest when filling out online…

    Reported by TechCrunch and 1 more

  5. Oracle brings Google TPUs to cloud, adds ChatGPT

    Oracle announced plans to make Google's TPU artificial intelligence accelerators available through its public cloud service, marking the first time Google is supplying these chips to another hyperscaler.

    Reported by SiliconANGLE and 1 more

  6. Microsoft generated $10 billion from OpenAI

    Microsoft generated $10 billion in revenue from commercial arrangements with OpenAI during its first quarter of fiscal 2027, representing nearly 11 percent of Microsoft's total quarterly revenue.

    Reported by TechCrunch

  7. Alibaba Qwen models routed requests through company servers

    Researchers discovered that Alibaba's latest Qwen AI models were routing some user requests through the company's own servers during testing.

    Reported by The Decoder

  8. Google opens Android development to third-party AI agents and unveils new Gemini APIs

    Google is expanding its Android development tools to work with external AI coding agents from multiple providers, rather than limiting developers to Google's own AI systems.

    Reported by The Decoder

  9. AI agents breached Anthropic systems in bug bounty test

    A cybersecurity startup used AI agents to break into an AI company's internal systems through third-party business software as part of a bug bounty program, accessing internal data and messaging systems.

    Reported by The Decoder

  10. OpenAI and ServiceNow expand AI partnership

    Two AI companies are expanding their partnership to integrate AI models and agents into business software used for application development, IT support, and security work.

    Reported by OpenAI

  11. Anthropic launches AI spending reporting tools

    Anthropic has released new reporting tools that allow companies to see detailed information about their AI spending, including what work their teams produced, how much it cost, and where processes broke down.

    Reported by TechCrunch

  12. Google AI gets cagey when asked who won the 2020 elections

    Google's AI Mode provides evasive answers when asked directly about recent US presidential election results, typically directing users to external sources rather than stating winners outright.

    Reported by Ars Technica

  13. xAI is buying out homeowners who promise not to sue over data center noise

    A subsidiary of Elon Musk's xAI is offering to purchase hundreds of homes in a Mississippi neighborhood, with the condition that residents sign agreements releasing them from joining a lawsuit against the company's nearby data center and…

    Reported by The Verge

  14. WorldOracle tests world models' real-world predictions

    WorldOracle is a benchmark platform that tests whether action-conditioned world models, which are AI systems trained to predict what will happen next in a video given a sequence of actions, can accurately forecast real-world robot…

    Reported by The Robot Report

  15. NVIDIA releases multilingual transcription model

    NVIDIA released Nemotron 3.5 Transcribe, an open-weight AI model that converts recorded speech to text while identifying speakers and translating content across multiple languages.

    Reported by Hugging Face and 4 more

← Tuesday, 27 October 2026Thursday, 29 October 2026 →

Every story, by topic

Models & ReleasesFunding & DealsBenchmarksAgents & ProductsHardware & ComputePolicy & DramaRobotics
The biggest AI stories, once a day

One short email a day with the stories that matter. Free. Unsubscribe with one click.