Syntheticauthenticity.

Synthetic authenticity is the permission people give an artificial persona to take on a role that matters while knowing it is artificial.

Prashant Saxena and Andrew Prahl formalised the threshold model of synthetic authenticity in the International Journal of Human-Computer Interaction in 2026. It names a familiar judgement. People can recognise a machine and still permit it to teach, advise, comfort, represent or act. The practical question is how far that permission should extend.

The Synthetic Authenticity framework.

Synthetic authenticity asks when an AI should be allowed to act in a role that matters. The framework helps leaders set limits, assign responsibility and protect trust as the AI changes.

What can this AI do, who is it acting for and who is responsible?

The more authority it has, the stronger the permission it needs.

  1. Inform
  2. Recommend
  3. Decide
  4. Transact
  5. Commit

Version threshold

A change to the model, memory, rules or owner can create a meaningfully different AI.

Can we identify the exact version that acted and what changed?

Legitimacy gapThe gap opens when an AI can do more than it has been given permission to do.

Review its permission when the model, memory, role, people it serves or stakes change.

Institutional trust grows when what AI can do, what it may do and who answers for it stay aligned.

The threshold model shows why more realism can become counterproductive. Human cues help until they invite closer scrutiny. The useful design point is calibrated sufficiency, where the persona carries the cues its role requires and no more.

Conceptual threshold model showing too few cues, calibrated sufficiency and closer scrutiny
Too little cueing reads as thin. Too much invites scrutiny. The target is the threshold between them.

Where AI should act, and where people must decide

The same question now reaches every institution. The framework makes clear what AI may do, what must remain human and what will earn the confidence of the people affected.

The decision Set the right boundaries for AI

Decide where AI may inform, persuade, teach, care or act, and where people must remain accountable.

Markets and brandsRepresentation, persuasion and purchasing

What may AI say, recommend or sell on an organisation’s behalf, and what must be disclosed?

OrganisationsDelegation, evidence and accountability

What may AI advise, decide or execute, and who remains answerable?

Public institutionsAuthority, explanation and recourse

What may AI do on behalf of the public, when must a person intervene and how can a decision be challenged?

EducationTeaching, assessment and authorship

What may AI teach, assess or produce, and where must faculty judgement remain?

Care and relationshipsDisclosure, dependence and dignity

What may AI recommend or invite, and where must human duty and dignity remain?

Platforms and cultureIdentity, influence and collective scrutiny

When AI shapes attention and influence, what must people know and be able to question?

From evidence to institutional decisions

The research programme examines role fit, disclosure, scrutiny, dependence and public judgement across AI personas. It gives boards, universities and public institutions a way to decide what AI may do, where human accountability must remain and what evidence or recourse people need to trust the result.

The wider vocabulary includes the transparency tax, crowd forensics, burdenless listening, scrutiny composition and ontological load.

How to cite the term Saxena, P., & Prahl, A. (2026). The authenticity paradox: The threshold model of synthetic authenticity. International Journal of Human-Computer Interaction. Advance online publication. https://doi.org/10.1080/10447318.2026.2680242
Concepts Research studies Practice Public writing