Deutsch English
← back to archive
AI & Technology · 645 words · MN-260630-0164

Russians and AI: Half the Fear, Full Overconfidence

MuckyNews reading

The Russian Knowledge Society ran a phone survey of 1,600 people to ask how they feel about AI-generated images, videos, and texts. The result is almost a perfect split: 47 percent are worried, 49 percent think it’s no big deal. But if you look past the raw numbers, the real story is that people are clearly unsure what to make of this, and their self-assessment is completely out of whack.

The biggest concern, at 29 percent, is straightforward fraud. Only 5 percent worry about political manipulation — which, coming from a country with a tightly managed state media landscape, isn't exactly shocking. When the information environment is already heavily curated, the idea of "manipulation" probably feels rather abstract to most respondents.

Things get really interesting on the confidence front. Over half — 51 percent — believe they can spot AI content when they see it. The youth are the most convinced: a solid 81 percent of the 18-to-24 crowd consider themselves human fake-detectors. Reality, however, begs to differ. 42 percent of respondents admitted they have already mistaken AI content for the real thing. A quarter of those had fallen for it more than once. The survey’s own punchline illustrates this perfectly: the "fluffsuckers." Back in early June, videos circulated showing little robots cleaning poplar fluff off the streets of Moscow. People thought they were real; they were AI. A harmless joke, sure, but it exposes the real pressure point: overconfidence is massive, but accuracy is remarkably poor. And as age goes up, actual ability to spot fakes drops, while the need for what the survey's organizers politely call "digital literacy" goes straight up.

Short version

  • What happened? A nationwide Russian survey shows the public is evenly

split on whether AI-generated content is a threat.

  • Who is involved? 1,600 respondents, the Russian Knowledge Society, and a

whole lot of internet users who think they can spot a fake but often can't.

  • Why does it matter? The gap between how good people think they are

at spotting AI content and how often they actually get fooled is huge — and it grows with age.

  • What is still open? Whether the 51 percent who feel confident in their

detection skills will ever realize they are mostly fooling themselves.

  • What is the real pressure point? Widespread overconfidence colliding

with active fraud risk. People don't just get fooled; they confidently walk right into it.

Perspective delta

russian_state The state news agency TASS frames the issue as a teachable moment for society. The survey is presented as proof that citizens are becoming aware of the challenges, and the narrative pushes digital literacy as the cure. The framing emphasizes individual gullibility and everyday fraud while sidelining structural concerns: only 5 percent see political manipulation as a threat. This aligns neatly with state interests. It positions the authorities as benevolent educators helping people navigate technology, while completely ignoring the possibility that state-aligned actors might themselves be the ones deploying AI-generated content for political purposes.

Source base / limits

The source base is extremely narrow. All material comes exclusively from TASS, representing the russian_state perspective. There are no independent or Western counterpoints, no civil society voices, and no critical scrutiny of the survey's methodology beyond the official margin of error. The raw text is compact, around 4,000 characters, providing only what the Knowledge Society and TASS chose to release. Whether respondents are genuinely unbothered by political manipulation or whether the survey's phrasing simply failed to capture that concern remains unclear. This is a single-perspective snapshot and should be read as exactly that.

Sources / basis

  • MuckyNews container ID: MN-20260630-164
  • Sources: TASS
  • Perspective groups: russian_state
  • Article count: 2
  • Source count: 1
  • Perspective count: 1
  • Approximate raw text: 4005 characters