Skip to content
IIAIT

AI Today · 15 November 2026

Nvidia names an interim CEO after Jensen Huang resigns, and Cloudflare open-sources its browser engine for AI agents

A quieter weekend brief: the leading AI chipmaker changes leadership, Cloudflare publishes the code of its lightweight browser for agents, ElevenLabs opens early access to a meeting and dictation app, and llama.cpp reaches version 1.0 with stricter defaults.

  1. 01

    Nvidia's CFO Colette Kress becomes interim CEO after Jensen Huang resigns

    Nvidia announced on November 12, 2026 that its board accepted the resignation of co-founder Jensen Huang as president, CEO and board member, effective immediately. Chief financial officer Colette Kress serves as interim CEO while keeping her finance role, and Harvey C. Jones becomes the company's first board chair. The board says it will consider internal and external candidates for a permanent CEO and expects operations, customer commitments and the product roadmap to continue without interruption. Fortune reports that the resignation followed Huang's admission of an affair that raised national security concerns.

    What it means for you

    Nothing changes today for the AI tools your team uses, since most of them run on Nvidia hardware through cloud providers that set their own prices. If you plan GPU purchases or cloud capacity, follow the earnings report on November 17 for any change to supply or roadmap guidance.

  2. 02

    Cloudflare releases Kitesurf, its experimental browser engine for AI agents, as open source

    Cloudflare published the code of Kitesurf on November 13, 2026 under the Apache 2.0 license. Kitesurf loads web pages, applies styles, runs page scripts and takes screenshots on Cloudflare Workers without running a full Chrome browser, and it accepts connections from automation tools such as Puppeteer. Developers can run it locally or deploy their own copy, which requires a paid Workers plan. Cloudflare says the project is still under heavy development, and Unite.AI notes it has no built-in authentication and is not yet suited to video or long signed-in sessions.

    What it means for you

    Teams building agents that read websites can test a lighter alternative to running full browsers, which may lower compute costs. If you deploy it, put an access control layer in front of it first, because a public endpoint could be used by anyone at your expense.

  3. 03

    ElevenLabs opens early access to ElevenAI, a Mac app for meeting notes and dictation

    ElevenLabs introduced ElevenAI at its New York summit on November 11, 2026 and opened early access the same day. The app takes notes in person or on calls without adding a bot to the meeting, produces action items and speaker labels, drafts follow-ups and offers dictation in any app in more than 90 languages. The company says it does not store audio and that transcript history stays on the user's Mac, and the app shows a notice on video when it is taking notes. ElevenLabs lists it as free for three months on Mac; TechCrunch describes it as an experimental app still in preview.

    What it means for you

    If you are comparing meeting assistants, this one is worth a trial for teams on Mac, especially those who prefer transcripts kept on the device. Whatever tool you use, tell all participants that notes are being taken and check your company's rules on recording before you start.

  4. 04

    llama.cpp reaches version 1.0 with a versioning policy and stricter server defaults

    The maintainers of llama.cpp, a widely used engine for running AI models on local hardware, released version 1.0.0 on November 13, 2026 together with ggml 1.0.0. They now follow semantic versioning, so breaking changes to the public interface will come only with a new major version. The release removes 36 deprecated functions, and its server now requires explicit opt-in flags for unauthenticated tool use and for fetching media from private network addresses, while the HF_TOKEN credential is sent only to huggingface.co.

    What it means for you

    If your team runs models locally with llama.cpp or tools built on it, test the upgrade before rolling it out, since older integrations and server setups may need changes. The stricter defaults are a reason to upgrade, because they reduce the risk of a local model server being misused.