In AI answers, Microsoft is the productivity and platform name, rarely the model builder.

This snapshot (29 September 2026) covers 250 benchmark questions put to ChatGPT, Claude, Gemini, Perplexity and Grok, or 1,250 answers in all. None of the questions named a brand. Microsoft AI appeared in 512 of those answers, roughly 41%. Grok (132 of 250) and ChatGPT (121) named it most often. Claude (75) and Perplexity (72) named it least.

The pattern by question intent is uneven. Microsoft AI's visibility was highest on leadership (69%), sustainability (68%) and developer ecosystem (66%). It was lowest on model capability, at 10%. In the answers, Microsoft shows up as the company that puts AI inside everyday software, sells it to governments and hospitals, and sits behind OpenAI. It is not framed as the builder of a leading model.

On productivity software, both ChatGPT and Gemini pointed to Microsoft 365 and Copilot. On foundation models, Perplexity separated distribution from model ownership. Its investor answer listed Microsoft and Amazon Web Services as platforms and distributors, not model leaders. Claude's answer on safety cited Microsoft's ISO 42001 certification. Gemini's reputation answer described a "split" profile, with praise for social initiatives and poor marks in AI ethics scorecards over defense and surveillance contracting.

Most mentions were classed as passing. In the credit tags recorded, the largest bucket was Not Substantive (14), followed by Sustainability & Ethics (4). Customer Experience and Safety & Compliance each had 2. Microsoft is often on the list without being singled out for one quality.

Peer rates need care because the question sets differ in size. OpenAI was named in 694 of 1,000 answers, Google DeepMind in 691 of 1,000 and Anthropic in 662 of 1,000.

PeakMetrics' news index logged 41,094 Microsoft mentions in the last 30 days. In 13,401 of them the brand was in the headline. The sampled headlines lean toward investors and enterprise deals. They include an analyst upgrade citing steady margins without heavy AI model spending, a Zacks stock-screen piece, a Copilot in PowerPoint webinar, and an expanded collaboration with HUMAIN at LEAP 2026 covering an enterprise AI offering and an AI PC.

For comparison, the index shows OpenAI at 155,993 mentions and Anthropic at 132,852. Google DeepMind sits at 43,848.

The press story is about financial performance, analyst views and enterprise rollouts. The answer story is about integration into workplace software, government and healthcare suitability, and sustainability claims. Both put Microsoft in the enterprise, but the press does not frame it around sustainability or leadership questions, where the answers named it most.

The largest difference is on models. Press coverage is heavy, yet on model-capability questions Microsoft AI was named in only 10% of answers. In the sampled headlines, one analyst piece even frames Microsoft's margins as holding up without heavy AI model spending. The answers make a similar split, treating Microsoft as the distributor and OpenAI as the model maker.

The answers also carried some critical framing. Gemini's worst-reputation answer named Microsoft alongside Meta, xAI and OpenAI. Claude's safety answers listed Microsoft in company lists, and Grok's promise-keeping answer included it among AI companies whose safety pledges independent assessments describe as weakened or conditional. None of the press headlines sampled address this.

This is one snapshot of live answers, and results vary between runs. It shows where Microsoft is placed today, not a trend.

A comms team would look first at the model-capability gap and ask whether the assistants are describing a deliberate positioning (platform and distributor) or an omission. It should also read the Gemini reputation answer and the Grok safety-pledge answer in full and see which sources they draw on. Sustainability and developer-ecosystem visibility are strong and worth confirming against cited domains in a wider sample.

Finally, compare press volume with answer visibility by intent, not in aggregate. High coverage did not translate into visibility on every question type.