AI Transparency Double Standard: Half of Adults Skip Disclosure

What happened

A new survey has found that roughly half of adults would publish work generated predominantly by artificial intelligence without disclosing that fact — even though the same respondents believe other people should disclose when they do exactly that. The findings, reported by TechRadar, expose a striking gap between the disclosure standards people hold others to and the ones they apply to themselves.

The research highlights what might be called an AI-transparency double standard: respondents broadly agreed that AI-generated content ought to be labelled as such, yet a near-majority said they would not follow that principle when it came to their own output — particularly when they had invested effort in crafting the prompt or refining the result.

Why it matters

For anyone designing customer-facing services, communications or content pipelines, this finding is directly consequential. Trust is the foundational currency of customer experience, and disclosure — or its absence — shapes how customers perceive authenticity, effort and accountability. If the people producing content, support responses or marketing copy operate under a private set of rules that differs from the public norm they endorse, the resulting asymmetry creates a latent credibility risk. When customers eventually discover undisclosed AI involvement (and the evidence suggests they often do), the damage to trust tends to be disproportionate to the original omission.

From a behavioural-economics perspective, the pattern is a textbook instance of self-serving bias compounded by the effort heuristic: people who invest time in prompt engineering begin to perceive the output as meaningfully "theirs," which lowers the psychological cost of non-disclosure. Service designers and CX leaders should treat this not as a moral failing but as a predictable cognitive tendency — one that organisational policy and transparent-by-default tooling can help correct before it becomes a brand liability.

By the numbers

• ~50% of adult respondents said they would publish predominantly AI-generated work without labelling it as such.
• A majority of the same respondents believe others should disclose when their work is mainly AI-generated — illustrating the gap between personal practice and the standard applied to peers.

The Renascence take

The instinct to frame this as a honesty problem misses the more actionable insight: this is a system design problem. Organisations that leave disclosure to individual conscience will get inconsistent, self-interested outcomes — because that is what the behavioural science predicts.

What most commentators overlook is that the effort heuristic makes non-disclosure feel justified, not dishonest — people genuinely believe a well-crafted prompt constitutes meaningful authorship. The fix is not an ethics lecture; it is friction-by-design. Embedding a mandatory disclosure toggle into publishing workflows, or making AI-assisted content the labelled default rather than the exception, shifts the cognitive load away from individual judgement and onto the system itself. Customer-obsessed operators should audit their content and service pipelines now — not to catch wrongdoing, but to close the gap between the transparency standard they publicly endorse and the one their processes actually enforce.