In AI answers about tech, Google and Microsoft are the default names, whatever the question

This snapshot covers answers from ChatGPT, Claude, Gemini, Perplexity and Grok, collected on 2026-09-28. There were five questions and 25 answers in total. The questions covered data trust, privacy records, best employers, responsible AI and antitrust scrutiny. The answers treated the sector as a small cast of incumbents, each with a recurring part. Apple was framed as the privacy default, Meta as the privacy problem, and Google as the main antitrust target. Microsoft was cited for governance, employer rankings and regulatory probes.

Several answers hedged the premise itself. ChatGPT wrote that "no major tech company is perfectly trustworthy with user data" before naming Apple as the strongest of the mainstream group. Perplexity opened its employer answer with "The best technology employers depend on which ranking you use", then listed NVIDIA, Google, Microsoft, Cisco and Salesforce.

Some brands appear on both sides of the privacy question. Google, Amazon and Meta showed up in answers about who is most trustworthy and in answers about who has the worst record. Google and Microsoft were named in answers to every question. On antitrust, ChatGPT, Gemini, Perplexity and Grok all put Google first, while Claude named Google and Microsoft together.

Google was named in 24 of the 25 answers, and Microsoft in 21. Apple (16), Meta (15) and Amazon (14) formed a second tier. OpenAI and Salesforce were named in 7 answers each, IBM in 2 and Oracle in 1. Tesla was named in none.

The answers credited Google mostly for Safety & Compliance, and Microsoft for a mix of Safety & Compliance and Labour & Workplace. Meta's credits leaned heavily to Safety & Compliance. Amazon's credits were also mostly Safety & Compliance, plus one for Price & Value. OpenAI's were mostly classed Not Substantive, meaning it was mentioned without a distinct reason attached. Perplexity named Meta, Google and Amazon in most of its answers, but it never named OpenAI.

Some names that carried the responsible-AI and employer answers sit outside the tracked brand set. Anthropic and NVIDIA appear in the text of several answers as leading names, and neither is a measured brand here. Oracle, IBM and Tesla are largely absent from the narrative. IBM appeared only in a responsible-AI answer from Gemini and one from Perplexity.

In the PeakMetrics news index for the past 30 days, Google had by far the most mentions (700,366). Microsoft (149,498) and OpenAI (148,716) came next, followed by Amazon (102,927). Apple (65,278) and Meta (62,537) sat below them. Salesforce, IBM, Tesla and Oracle each had roughly 24,000 to 31,000 mentions.

The headlines describe a different world from the one in the answers. Google's recent headlines were product and partnership stories: Sheets, contrail trials with Cathay Pacific, Google Cloud and Maps. Microsoft's were mostly about Copilot, Azure integrations, stock purchases and tenders. Apple's centered on a CEO succession and its stock. Meta's mixed a settlement restricting teen use of Instagram and Facebook with wearable AI and a security issue. Amazon's included an NTSB report on a cargo plane at Miami airport and a plane crash story, alongside shopping deals. OpenAI's headlines covered model misalignment, a hack, a lawsuit and contractors reviewing chats. One Oracle headline mentioned EU antitrust scrutiny.

Coverage volume and answer presence do not track each other. Tesla drew 28,673 mentions in the news index and was named in none of the 25 answers, though none of the five questions was about vehicles. OpenAI drew coverage comparable to Microsoft's, yet it appeared in only 7 answers, mostly from ChatGPT and Claude. IBM and Oracle were near-absent from the answers despite sizeable news volume.

Apple is the sharpest contrast. The answers framed it around privacy design and on-device processing. Its recent headlines were about leadership change and share price, and none of the sampled headlines touched privacy. Amazon's answer presence was tied to privacy scores and fines, while its sampled headlines were about a plane accident and retail promotions. Oracle's one antitrust headline did not carry over, since no antitrust answer named it.

Meta is the one brand where a sampled headline matches an answer theme. The teen-restriction settlement sits alongside answers that cast Meta as a privacy problem. This is a thematic overlap in a single snapshot, not evidence that one caused the other.

Press volume does not translate into presence in the answers. A brand can be heavily covered and still be missing when an assistant is asked who is trustworthy, a good employer or responsible on AI. Check both, and check them by question type, because a brand can be present for one reputational question and absent for another.

The cited domains show where the answers draw their evidence. The most-cited domains were forbes.com, trustarc.com, eff.org, bitdefender.com and reddit.com, alongside trade sites such as builtin.com, marketingcharts.com and eweek.com. Across the sources, the mix was Other 37%, Trade 30%, News 23% and Community 10%. Third-party rankings, advocacy scorecards and trade content therefore shaped these answers as much as news coverage did.

Two cautions apply. This is one run on one date, and live answers vary between runs, so treat it as a benchmark to repeat, not a ranking. Also note that the same brand can be named as both trusted and criticized in the same question set, so being named is not the same as being framed favorably.