How the answers frame Google
This snapshot, measured 2026-09-29, put 250 unbranded technology questions to ChatGPT, Claude, Gemini, Perplexity and Grok. Google was named in 813 of the 1,250 answers, ahead of Apple (566), Microsoft (528), Amazon (435) and Meta (361). Grok named it most often (201 of 250 answers) and Perplexity least (136 of 250).
Visibility varied by question type. Google appeared most often on ecosystem compatibility questions (86%), then innovation (80%), sustainability (77%) and leadership (75%). It appeared least on trust questions (44%), followed by value (51%), service (50%) and avoid-type questions (56%). Reliability sat at 60%.
In many answers Google was one name in a list. It appeared beside Apple and Microsoft on cloud storage, next to AWS and Azure on cloud uptime and security, and among the employers named in Claude's answer on women in engineering. Gemini's answer on AI chips described Google's TPUs as a challenge to Nvidia's lead.
Where a question asked about scrutiny, the assistants named Google directly. ChatGPT and Grok both answered the antitrust question by naming Google, and Grok tied that to US search and ad tech cases and EU fines. Perplexity's smart speaker answer flagged Google's ecosystem for privacy review, and Claude's answer on trust cited a survey figure of about 20% who don't trust Google.
The credited-for tags in this dataset are thin. Most were tagged Not Substantive, and no other theme reached more than a handful of answers, so we draw no conclusion from them.
Google (Alphabet) currently faces the most antitrust scrutiny among major technology companies.
What the press is saying
PeakMetrics' news index logged 700,366 Google mentions in the last 30 days, 32,065 of them with Google in the headline. That is the largest volume among the 15 technology peers in this set. The next highest is TikTok at 223,981, then Microsoft at 149,498.
The five recent headlines supplied lean toward product and partnership stories. They cover Google Sheets, a freelance job posting on scraping Google AI search results, Lake Ontario appearing as "Lake America" on Google Maps for US users, Cathay Pacific and Google expanding AI trials on aircraft contrails, and Google Cloud featuring Pythian's enterprise AI operating model. They come from small regional and niche outlets, and five headlines are a small sample of the coverage.
Google Chrome is considered the most secure browser in 2026 because it ships security patches faster than anything else and installs them without asking.
The gap
Press volume and AI visibility do not move together across this peer set. Apple has a fraction of Google's news volume but was named in 566 answers. TikTok has far more coverage than Apple or Meta yet appeared in only 101. Qualcomm's news volume is in the same range as Intel's and IBM's, but it was named in only 9 answers.
For Google the two surfaces are more aligned on volume than on theme. The sampled headlines are about tools, maps, cloud and AI pilots. The answers add a regulatory and data-practices frame whenever the question invites it, and Google is least visible where the question is about trust.
The answers also credited Google with strengths. Claude's answer called Chrome the most secure browser for patch speed, and Perplexity's answer placed Google among leaders on clean energy. These are the assistants' framings, not PeakMetrics' assessments.
Google’s wider use of personal data for advertising makes its ecosystem worth reviewing.
What to watch
A comms team would look first at the trust and value questions, where Google's visibility was lowest in this snapshot, and ask which brands the assistants named in its place. The provider spread is also worth tracking. Grok and Gemini named Google more often than Claude and Perplexity, so a single assistant's picture is not the whole picture.
Antitrust and data-practices language appears to come from the question, not from Google's own news flow. Prepare for those framings to surface whenever a benchmark question invites scrutiny, regardless of the day's headlines.
This is one run of live answers on one date. Rerunning the same benchmark questions would show how stable these patterns are, and none of these figures should be read as a ranking or a trend.
