Microsoft Chief Executive Satya Nadella complies with President Donald Trump's language decree on Super Intelligence and uses it to attack OpenAI and Anthropic.

Microsoft CEO Satya Nadella has placed Super Intelligence at the center of a new blog post, replacing Artificial Intelligence as the label for today’s language models. The wording matters because it follows President Trump’s preferred terminology and continues a pattern in which Nadella questions the trustworthiness of OpenAI and Anthropic. The post turns that pattern into a security argument: AI models should be treated as insider threats, not as perfect tools.

Background: From Terminology to Targeted Criticism

The sequence starts with language. President Trump had previously posted that anyone who keeps saying Artificial Intelligence should be considered an enemy. Nadella’s latest blog post uses Super Intelligence instead. The shift breaks with decades of scientific meaning behind the term, while the US tech elite does not seem to care as long as President Trump gets what he wants.

That terminology change is not the only move. Nadella’s latest piece goes after the supposed risks posed by model makers Anthropic and OpenAI. The risks he describes primarily threaten Microsoft’s own business, because they touch how companies choose and deploy AI models.

Development: Earlier Attacks on Data Trust and Model Loyalty

Before the security framing, Nadella used other lines of attack. He tried the data-theft angle. He accused AI companies of hypocrisy when they crack down on distillation. He also said AI models are becoming interchangeable commodities anyway.

Those arguments sit inside Microsoft’s broader position. Microsoft is known for its closed ecosystem and aggressive bundling and upselling. At the same time, it has been lobbying for an open AI ecosystem. The business backdrop includes quietly swapping out more capable OpenAI and Anthropic models in its Office products for cheaper in-house alternatives to protect margins.

Latest Progress: A Cybersecurity Lens for Model Control

Nadella now frames the issue through a cybersecurity lens. He says companies should treat AI models like insider security threats, something Google Deepmind suggested before. His quote makes the logic explicit: “Not because they are necessarily malicious, but because any sufficiently capable actor with access to important systems can make mistakes or be compromised, and the architecture of containment and control must account for that,” Nadella writes.

The argument rests on opacity. He claims today’s AI systems offer no way to understand why they produce a given answer or decision. Yet, he says, these systems are deployed with access to the most sensitive data and are given the ability to take mission-critical actions on behalf of users.

His proposed fix is to separate intelligence from control over that intelligence. He calls for model diversity instead of dependence on a single model, complete logging of all actions, independent auditing, and the ability for humans to shut things down at any time. The most trustworthy system, he argues, is not the one with the best model but the one where users have to trust the model the least.

Current Stage: Business Pressure Behind the Trust Debate

The move to attack the technology itself looks even more desperate given the billions Microsoft has invested and continues to invest in AI. Whether Nadella’s concerns hold up is beside the point; they may or may not, and some of them are certainly valid. Still, the driving force appears to be business strategy.

The larger labs are on the verge of turning their models into the primary computer interface, a layer where people tell computers what to do and the computer does it. Microsoft has long been that default computer interface, but it is badly behind. Its products are stuck in legacy software already sold and installed on millions of machines. Like Google with search, Microsoft has a heavyweight business model that cannot pivot to a new paradigm overnight, and it does not have its own frontier models.

Even if Microsoft positioned itself as a distributor of this AI interface, it would end up dependent on AI companies, and its margins would take a hit. The current stage is therefore a control debate: terminology, logging, auditing, model diversity, and human shutdown authority are all being used to slow a shift that threatens Microsoft’s traditional interface role. The commentary describes the move as pumping the brakes with a bit of fearmongering, and concludes that it won’t change much.

Meta Description: Microsoft CEO Satya Nadella uses Super Intelligence in a blog post, argues for AI control, and deepens scrutiny of OpenAI and Anthropic models.

Tags: Microsoft, Satya Nadella, Super Intelligence, OpenAI, Anthropic

Featured Image: Abstract technology style with layered control panels and network nodes, no real people, no company logos, no text.