Gallup counts more than two Americans distrusting AI for every one who trusts it

Confined to people who hold a view either way, "more than twice as many" distrusting as trusting puts distrust at no less than two-thirds of that group ^[13]. Vox's account gives the ratio without the two underlying percentages, so the share of the full sample cannot be recovered from it. The moving number in the same survey is the one closest to a deployment case: a growing share of respondents expect AI to have a negative effect on productivity and economic growth ^[2].

Set that ratio ^[1] beside Pew's finding that about half of Americans use chatbots ^[3] and the two measures are describing different things. Seat activation on an internal dashboard can climb while the same people hold the view Gallup recorded.

The cause named in the reporting is procedural. "It's being forced on users, there's no clear way to turn it off, and it feels like this unwelcome addition to their lives," Rory Mir of the Electronic Frontier Foundation, a nonprofit that defends digital privacy and free speech, told Vox ^{[6] [15]}. Mir said the pushback comes in part from "this kind of continued pattern of disrespect to users" ^[7].

None of this is measured at the office door. The Vox account reports attitudes and consumer usage among Americans and does not report any workplace-deployment measure ^[8]. Annoyance at AI in a search result predicts nothing about whether a claims team will use an internal assistant. The delivery path is where the two connect: it runs through the same products people use at work. "More and more AI use is through its integration into digital products like search, email, camera apps, and social media," Arvind Narayanan, a computer science professor at Princeton University, told Vox ^{[9] [16]}, adding that it "will become increasingly hard to strip these out unless one is willing to forgo digital products altogether" ^[10]. (Vox disclosed that Vox Media is one of several publishers that have signed partnership agreements with OpenAI ^[17].)

There is also a recent precedent for the data answer arriving after the fact. Meta told some Facebook and Instagram users in 2024 that it was using their public photos and text to train its AI models. That prompted global outrage, and Meta still does it ^[12].

So the tradeoff is between the speed of uptake and the terms of uptake. Switch the feature on for everyone and the numbers arrive this quarter, along with the resistance Mir describes. Publish the off switch and the training-data path first and the numbers come more slowly. Vox's guide says turning AI features off is possible but tedious, and that keeping personal data away from web scrapers is harder still ^[11]. Inside a company that tedium has an owner, usually whoever administers the suite, and Pew's majority expecting AI to put their personal information at risk is the objection an employee-facing rollout has to answer ^[4].