Launching the largest artificial intelligence experiment ever


Good morning {{first_name| Artificial intelligence lovers}}. Finally, the two faces behind the biggest AI competition are in the same room – but this time it’s a federal courtroom, with Elon Musk on the witness stand and Sam Altman watching from the gallery.

Musk’s $130 billion trial against OpenAI began on Tuesday, with four weeks of testimony, hundreds of pages of private emails set to leak into the public record, major names in AI on the stand, and plenty of drama still to come.

  • Elon Musk’s $130 Billion Trial Against OpenAI Begins

  • Google finalizes Pentagon deal despite protests

  • Automate any manual task with Codex

  • Talkie is an artificial intelligence that thinks it’s the year 1930

  • 4 new tools for AI, community workflows, and more

Obinay

Image source: Images 2.0 / Rundown

Rundown: Only Elon Musk Take the stand in federal court as opening statements began in his $130 billion lawsuit against OpenAI, accusing CEO Sam Altman of “stealing a charity” – while OAI’s lawyer told the jury that Musk sued because he “didn’t get what he wanted.”

  • Musk’s lawsuit seeks $130 billion in damages, the removal of Altman and Brockman from the board, and the forced reversal of OpenAI’s recent for-profit conversion.

  • “If a ruling is made that it is okay to loot a charity, the entire foundation of charitable giving in America will be damaged,” Musk testified.

  • OAI’s legal team called Musk’s lawsuit “sour grapes,” saying Musk didn’t like the success the company saw after his departure.

  • Microsoft’s legal team said Musk did not object to OAI’s structure until after its success as a competitor to xAI, and said he “knew nothing” about Altman being fired in 2023.

Why it matters: This is just the first day of one of the most controversial court cases the tech world has ever seen, and the details will be exciting. With high-profile AI figures set to testify and with hundreds of pages of private messages about to leak into the public record, the next four weeks will be hard to overlook.

Combined with teleportation

Rundown Running agents in production does not mean trading security for speed. Teleport Beams provides an isolated virtual machine and sets identity via a short-lived certificate before the agent runs a single line – no secrets, no IAM wrestling, and no perpetual privileges.

Designed for teams that ship real agents:

  • Each beam runs in an isolated Firecracker VM

  • Supports coding proxies, sandbox applications, and proxy functions

  • Fully auditable, with identity and access tied to each session

Google

Image source: Images 2.0 / Rundown

Rundown: Google I fell Secret AI deal with the Pentagon, opening its models to “any lawful government purpose,” the same week that more than 600 employees wrote an open letter to CEO Sundar Pichai, calling for a rejection of the use of AI for military purposes.

  • More than 600 employees at Google sender Pichai sent a letter on Monday asking him to “refuse to make our AI systems available for classified workloads.”

  • The information stated that the contract opens Google’s artificial intelligence to “any lawful government purpose,” with no legal right to object to how the Pentagon uses it.

  • OAI and xAI are signed Deals With the Pentagon last month, and with Anthropy currently Fighting In court after being blacklisted for not dropping his guardrails.

  • Google’s no-guns pledge has been revoked Principles of artificial intelligence In 2025, after it was implemented in 2018 after its success Employee protests.

Why it matters: The Pentagon drama may still seem fresh in the rivalry between OAI and Anthropic, but it’s not discouraging another AI lab from making a similar deal. Google is now wading into messy territory from a PR and internal perspective, and time will tell if the same backlash we saw with ChatGPT will now come to Gemini’s doorstep.

Artificial intelligence training

Rundown: In this guide, you’ll learn how to let OpenAI’s Codex click through any annoying, repetitive work using your computer on Mac or Windows.

  1. Open Codex, go to Plugins, find and enable the Computer Use plugin, and then start a new task

  2. Open the permissions menu and switch from default permissions to full access. Confirm any claims

  3. Give a real task, such as: “Open Chrome and debug the UI of this web page http://localhost:3000/. Click, reproduce the error I describe, and then tell me what you think is causing it. If you are not sure, ask before making changes.”

Pro Tip: Codex can automate repetitive workflows in on-premises applications as well. Try it with Photoshop exports, Premiere cleanup, file renaming, or any other tool.

Provided by TELY AI

Rundown: Your buyers ask AI questions, and your competitors’ AI answers them, not you. Tely lets AIs like ChatGPT, Google, and Claude recommend your business instead.

  • Get recommendations on ChatGPT, Google, Perplexity, and Cloud in less than one week

  • Completely hands-off: no writers, no agencies, and no content moderation

  • Lower costs than hiring freelancers or maintaining a marketing team

  • Ideal for specialized industries where expertise matters

Artificial intelligence research

Image source: Images 2.0 / Rundown

Rundown: Researchers Nick Levine, David Duvenaud (Fmr. Anthropic), and Alec Radford (Fmr. OpenAI) experimental Talkie, a “vintage” 13B AI model trained only on text before 1931, was designed to test how AI would think when its worldview predates the Internet.

  • Talkie was trained on 260 billion codes from pre-1931 books, newspapers, magazines, patents, and case law, all of which are now in the public domain of the United States.

  • To teach a native speaker how to chat without recent data, the team pulled instructions from etiquette manuals and cookbooks, with Claude Sonnet 4.6 sorting out the answers.

  • The programming language Python didn’t exist in 1930, but Talkie wrote working code by flipping the plus sign to a minus sign in one example, proving that it can be generalized.

  • AI benchmarks take a hit when models are trained on their own test data, Speech avoids this, and a GPT-3 level version comes next.

Why it matters: All current border models look vaguely similar because they read roughly the same as the modern web. The Talkie is certainly cut from a different cloth, but the Python coding tale is a fascinating part of the experience that demonstrates the kind of learning and thinking that is likely to occur under the original training data.

  • 📊 Replay the slide show – Create polished, stunning slides in seconds using AI

  • 🌎 Echo-2 – New global SOTA model for text to 3D conversion from SpAItial

  • ⚙️ Workflow – Mistral Foundation tool for sequencing artificial intelligence agents

OpenAI Announce GPT-5.5, Codex, and Managed Proxies are now available through Amazon Bedrock, one day after its new contract Restructuring With Microsoft.

Nvidia Released Nemotron 3 Nano Omni, a new open model that can handle vision, audio and text at 9 times the speed of competing open multimedia models.

Wall Street Journal I mentioned That OAI fell short of its targets for revenue and user growth, with chief financial officer Sarah Friar questioning its massive spending – OAI calling it “ridiculous”.

Anthropic He added New connectors for a wide range of creative workflows, including apps like Blender, Adobe Creative Cloud, Autodesk Fusion, SketchUp, and more.

Xiaomi Open source MiMo-V2.5-Pro, which ties Kimi K2.6 in the synthetic analysis leaderboard, features a 1M context window and powerful proxies for proxies.

My place Fired Echo-2, SOTA’s new world model that turns text or images into explorable 3D worlds, claims to outperform World Labs’ Marble 1.1 across benchmarks.

In each newsletter we show how the reader is using AI to work smarter, save time or make life easier.

Today’s workflow comes from the reader Alan C. In Ontario, Canada:

“I used AI to resolve what sounded like a furnace malfunction during a cold night without calling a technician. After installing the Google Nest Thermostat about a year ago, my system started making a loud “accelerating” noise and flashing error codes (2 and 7) indicating pressure switch or hardware errors.

I photographed the control board and wiring and uploaded it to Google Gemini. I located the furnace, traced the problem to the thermostat power theft, and reported a wiring error. She then ran a simple “G-to-C” wiring conversion to provide stable power, along with steps to remove lockouts and check the readings.

Within an hour, the interference stopped, the error codes were cleared, and the system was stable and operating normally. It saved me about $300 on an HVAC tech service call, and I avoided a night in the cold.”

How do you use artificial intelligence? Tell us here.

That’s all for today!

Before you go, we’d love to know what you thought of today’s newsletter to help us improve The Rundown experience for you.

Rowan, Joey, Zack, Shubham and Jennifer – the humans behind The Rundown

Leave a Reply