Microsoft's Nadella bows to Trump's language diktat on "Super Intelligence" and uses it to attack OpenAI and Anthropic

What happened

Microsoft CEO Satya Nadella has adopted the term "superintelligence" in place of "artificial intelligence" in his latest public commentary, aligning his language with terminology favoured by the Trump administration. In the same post, Nadella framed advanced AI models as a potential insider security risk, a characterisation that reads as a pointed critique of rivals OpenAI and Anthropic.

According to reporting by The Decoder, the shift in vocabulary coincides with a broader strategic concern at Microsoft: as AI model providers increasingly position their systems as the primary interface through which users interact with computers, Microsoft risks being pushed into a secondary role in the technology stack it has historically controlled through Windows and Office.

Why it matters

The episode illustrates how language itself has become a competitive and political instrument in the AI industry. Terminology choices — "AI" versus "superintelligence" — are no longer neutral technical descriptors; they signal alignment with policy narratives and can be used to reframe competitors' products as risks rather than innovations.

For enterprise and public-sector leaders evaluating AI vendors, this points to a market where messaging around safety, security and capability is increasingly shaped by commercial positioning as much as by technical substance. Decision-makers should look past vendor framing to assess actual model governance, security architecture and interoperability.

The Renascence take

Framing a competitor's product as an "insider threat" is a classic trust-shifting tactic: it redirects scrutiny away from one's own platform and onto rivals, while new terminology signals allegiance rather than adding clarity for the end user.

Language is never incidental in experience design — it sets expectations and shapes trust before a single feature is evaluated. When vendors compete on framing rather than on demonstrable security practices, buyers should insist on independent verification of how models are governed, tested and monitored, rather than accepting either "superintelligence" branding or insider-threat narratives at face value. The operators who win here will be those who can show, not just say, how their AI systems protect users.