OpenAI, Microsoft and AGI

Dear Readers,

Imagine there was a document that not only attempted to measure the transition to artificial general intelligence (AGI), but also decided who would be allowed to participate once it had been achieved. This is not a sci-fi script, but reality: OpenAI's internal paper on the “Five Levels of General AI Capabilities” is perhaps the most consequential text in recent AI history – precisely because it was not published.

The debate between OpenAI and Microsoft shows that it is no longer just about technology, but about institutional power relations, access rights, and economic hegemony in the age of superintelligence. When AGI becomes contractually relevant rather than hypothetical, the rules of the game shift. And not gradually, but eruptively.

In Today’s Issue:

All the best,

“A hidden nail is currently driving into Silicon Valley's most highly praised tech partnership – and this could be a turning point for the entire AI community.”

OpenAI and Microsoft appear to be at a critical crossroads. At the center of it all is a previously unpublished internal OpenAI document titled “Five Levels of General AI Capabilities.” It attempts to systematically describe when AI can be considered “general” – i.e., when it becomes more economically competent than humans in a wide range of tasks. But it is precisely this classification that could trigger a mechanism in the contracts with Microsoft that would deny the software giant access to future AGI technology. The paper was therefore not published despite being completed in May 2024, out of concern that it could inadvertently trigger contractual consequences.

This is explosive for the AI community, as it is the first time that information has leaked out about how the threshold to AGI is conceived institutionally and described technically. This shows that we are no longer just talking hypothetically about AGI – we are negotiating its political and economic consequences in real time.

What happens when a scientific definition becomes a billion-dollar turning point? The answer to this question will determine how transparent, democratic, and accessible our AI future will be.

Why it matters: This document could be the key to making AGI measurable and contractually tangible – or it could become the basis for opacity and power struggles in the AI race.

Sources:

🔗 https://www.wired.com/story/openai-five-levels-agi-paper-microsoft-negotiations/

🔗 https://www.wsj.com/tech/ai/openai-microsoft-rift-hinges-on-how-smart-ai-can-get-82566509

Unlock a focused set of AI strategies built to streamline your work and maximize impact. This guide delivers the practical tactics and tools marketers need to start seeing results right away:

Stay ahead of the curve with these top strategies AI helped develop for marketers, built for real-world results.

Download the Free Report

Nvidia CEO Jensen Huang argues that achieving greatness requires embracing "pain and suffering," as the struggle to build something difficult is more important than the pursuit of constant happiness.

The world's first fully autonomous soccer match between humanoid AI robots took place in China – three against three, without remote control. The robots analyze the game in real time, make tactical decisions, celebrate goals, and adapt to unforeseen events. The event shows how far networked sensor technology, visual AI, and motion control have advanced in humanoid robots – and opens new doors for robotics in team environments.

Researchers at MIT CSAIL are using generative AI coupled with physical simulation to design robot motors that enable 41% higher jumps and safe landings. A brain design optimized by GenAI significantly raises the performance limit of conventional designs. This shows how AI can fundamentally transform not only software but also hardware design – and opens up prospects for agile, self-designing robots in rescue, exploration, and manufacturing.

Nvidia and Foxconn are negotiating the use of humanoid robots at Foxconn's new AI server factory in Houston – a pilot for the first Nvidia product that will one day be manufactured by robots. The humanoid assistants, already tested in cable management and assembly, are expected to be operational from Q1 2026. This initiative marks a historic step: human production lines could soon be supplemented or replaced by humanoid robots – a disruptive change in industrial automation.