AI Transparency

• Organisations treat AI disclosure as a compliance box, not a trust‑design problem.
• Research shows disclosure reduces trust; hiding AI use and being caught reduces it even more.
• The mechanism is legitimacy: people expect human effort in tasks like writing, analysis, and decision support.
• Without legitimacy, even high‑quality AI output triggers social penalties and trust erosion.

• People judge *whether* AI use is socially appropriate for the task.
• When legitimacy is absent, disclosure becomes a liability.
• Mandated disclosure quietly taxes honest employees and rewards silent ones.
• This creates a behavioural loop: compliance → penalty; silence → reward.

• “Used AI” is an empty phrase: it reveals nothing about what AI enabled.
• Key missing variables: - what the model actually did, - how results were verified, - who holds accountability for conclusions.
• AI can analyse volumes of material beyond human capacity, altering sample size and generalisability.
• Trust penalties may shift depending on whether people understand the AI’s role and the verification chain.

• Transparency must be designed as part of consequence architecture, not paperwork.
• Policies should stabilise trust, not destabilise it.
• Leadership must establish norms before policy; legitimacy is a cultural signal, not a compliance form.

AI transparency is not about revealing tools; it is about designing trust.

• Disclosure without legitimacy or clarity creates governance drift.
• The correct question is: does your disclosure policy reward honesty or tax it?