AI labels made one video seem less authentic—but did not change trust
In a randomized online experiment with 617 young adults, an AI disclosure label slightly reduced perceived authenticity. It did not directly change trust, sharing, adoption intention or perceived risk.
The short version
- What happened
- Researchers randomized 617 adults aged 18–25 to one of four versions of the same AI-generated Pomodoro-method video: higher or lower quality, with or without an AI label.
- Who it affects
- Viewers may notice that content is synthetic without automatically trusting or rejecting it; informative context may matter more than a bare label.
- What remains unproven
- All participants were aged 18–25, and the study used one AI-generated video about the Pomodoro study method.
A label changed authenticity, not behaviour
A disclosure label made one AI-generated instructional video feel slightly less authentic, but it did not directly alter trust or intended behaviour. Researchers recruited 640 people through an online platform and retained 617 adults aged 18–25 after prespecified quality checks. Participants were randomly allocated to four versions of a Pomodoro-method video crossing higher or lower visual quality with the presence or absence of an AI label.[1]
Who it affects
Viewers may notice that content is synthetic without automatically trusting or rejecting it; informative context may matter more than a bare label.[1]
Keep in mind
All participants were aged 18–25, and the study used one AI-generated video about the Pomodoro study method.
Trust, risk and intentions were immediate self-reports; actual sharing, adoption and later recall were not measured.
Eight participants were excluded after randomisation because they did not recognise the assigned manipulation.
The authenticity effect was small, and non-significant direct effects do not establish that labels never affect trust or behaviour.
The methods and limitations sections describe stimulus validation differently, which warrants clarification.
Source analysis · 1 linked source. Original source date: 20 October 2026.
Keep following the evidence
Join The Impact Brief founding list for selected AI reports, their effect on people and the evidence limits. Read a sample first.
Read the full report · 3 min The evidence, methods, implications and limitations
A label changed authenticity, not behaviour
A disclosure label made one AI-generated instructional video feel slightly less authentic, but it did not directly alter trust or intended behaviour. Researchers recruited 640 people through an online platform and retained 617 adults aged 18–25 after prespecified quality checks. Participants were randomly allocated to four versions of a Pomodoro-method video crossing higher or lower visual quality with the presence or absence of an AI label.
The label's estimated effect on perceived authenticity was statistically significant but small: B −0.214, 95% confidence interval −0.406 to −0.022, with partial eta-squared 0.008. Direct effects on trust, adoption intention, sharing intention and perceived risk were not statistically significant. Video quality and the quality-by-label interaction were also non-significant for those main outcomes.[1]
Transparency is not a trust switch
The study's wider model linked authenticity, information-processing style, perceived risk and trust, but those variables were measured together immediately after viewing. That supports association, not a proven psychological sequence. Eight responses were also excluded for failing to identify the assigned manipulation after randomisation, which can affect the protection random allocation provides.
Generalisability is narrow: one neutral study-skills video, one short exposure and young-adult self-reports rather than observed sharing. The paper describes a 60-person stimulus pretest, while its limitations section says no separate independent validation study was conducted; that wording needs clarification. Platform labels should therefore be tested by content type, wording, placement and real behaviour rather than assumed to increase trust automatically.[1]
The distinction that matters
What changed after the AI label?
The randomized comparison found one direct effect and several non-effects.
Authenticity
AI label versus no label
Small reduction; p=0.029
Trust
AI label versus no label
No significant direct effect
Intentions
Adopt or share
No significant direct effect
Watch the background, then check the evidence
Illustrative photo · not a video frame
The Royal Society · 44 min 28 sec · Background explainer
Is this the AI future we want? | The Royal Society
YouTube loads only when you choose; playback starts when you press play.
A Royal Society discussion with Madhumita Murgia and biomedical engineer Alison Noble on AI’s potential in healthcare, science and culture, recorded at the 30 June 2026 event.
Evidence check: A discussion of opportunities and risks, including future possibilities. It is not evidence that every application described is available or clinically proven.
Published 2026-07-01. Source checked 14 October 2026. Check captions on the publisher’s video page.
Check the healthcare evidencePhoto: Daniel Sone / National Cancer Institute / Unsplash · Licence
Evidence check
All participants were aged 18–25, and the study used one AI-generated video about the Pomodoro study method.
Trust, risk and intentions were immediate self-reports; actual sharing, adoption and later recall were not measured.
Eight participants were excluded after randomisation because they did not recognise the assigned manipulation.
The authenticity effect was small, and non-significant direct effects do not establish that labels never affect trust or behaviour.
The methods and limitations sections describe stimulus validation differently, which warrants clarification.
Reporting basis: Source analysis. Original source date: 20 October 2026.
Continue the story
Read next
Continue exploring synthetic media: compare the finding and its evidence limits.
Can people spot AI-generated images?
A preregistered experiment with 2,091 adults in Denmark found that people were slightly more accurate than random probability guesses, yet on average would have scored better by assigning every image a 50% chance of being AI-generated. The test used one 2024 image tool and does not measure belief, sharing or harm.
Read the next reportExplore the full source and evidence record
What this means for people
- Viewers may notice that content is synthetic without automatically trusting or rejecting it; informative context may matter more than a bare label.
- Platforms and publishers should evaluate labels with real behaviour and high-stakes content rather than treating transparency as a complete safety intervention.
Global context
Disclosure rules and platform designs vary internationally. This study does not compare legal label formats or show how people respond to election, health, advertising or crisis content, where prior beliefs and stakes may be different.
What the evidence does not yet show
- All participants were aged 18–25, and the study used one AI-generated video about the Pomodoro study method.
- Trust, risk and intentions were immediate self-reports; actual sharing, adoption and later recall were not measured.
- Eight participants were excluded after randomisation because they did not recognise the assigned manipulation.
- The authenticity effect was small, and non-significant direct effects do not establish that labels never affect trust or behaviour.
- The methods and limitations sections describe stimulus validation differently, which warrants clarification.
What to watch next
- Pre-registered replications with varied ages, countries and media-literacy levels
- Comparisons of label wording, placement, source detail and explanation
- Observed sharing and reliance rather than only post-exposure intentions
- Tests using political, health, advertising and misinformation scenarios
Living evidence record
Impact record IAI-1WJRWPQ
Evidence stage
Studied
Confidence
Supported
Reporting basis
Source analysis
Independent or research support
Present
Record status
Monitoring
Last checked
21 October 2026
Source trail
1 direct source across 1 source type.
People impact
Documented in this record.
Uncertainty
Limits and next checks are explicit.
Stages describe the evidence available—not whether a technology is good or bad. See the public method.
Single-source reporting disclosure
This record analyses one direct source. It can establish what Frontiers in Psychology published or reported, but it is not independent corroboration of every performance claim or predicted outcome. The confidence label will change only when broader evidence is added.
Evidence trail
Sources used for this report
Links checked 21 October 2026
Research question, topics and regions
Whether disclosure labels and video quality change how young adults judge an AI-generated instructional video
Topics: Synthetic media · AI labels · Trust · Media literacy · Transparency
Regions: China · Asia
This report is labelled source analysis. We summarise and analyse source material in our own words; company statements remain attributed claims until independently supported. Translated summaries preserve the meaning of the original source and link back to it. Read our editorial standards.
Reader commentary
What does this mean where you are?
Tell us what you have seen at work, in your community or in everyday life. A question is welcome too. No account is needed; we review notes before publishing them.
Explore commentary across the portal →Published reader notes
0No published reader notes yet. You can start the evidence-led discussion above.
Prefer a private correction or response? Contact the newsroom.