The ethics of AI fortune-telling

The category of “AI fortune-telling apps” — palmistry, astrology, tarot, and adjacent symbolic readings delivered through phones with AI components — is growing rapidly, and the ethical floor of the category is being set right now by the apps that get built. The honest version of how to build these apps well is short. The dishonest version is more common, and worth naming directly.

What follows is the ethical framework PalmAura is built around. It is also offered as an open argument to other builders working in the same category. The principles are not proprietary. The standards should be common.

The real risks of AI fortune-telling

The ethical risks specific to AI fortune-telling are not the same as the risks of older fortune-telling formats. Four of them matter most.

Overpromising prediction. An AI palm reading app that claims to forecast events — when you will marry, when you will be wealthy, what disease you will avoid — is doing the single most damaging thing in the category. The claims are false (palmistry does not predict events, and no AI technique can change that), and the false claims affect users who treat them seriously. People do real things on the basis of fortune-telling app readings: leave relationships, take or refuse jobs, postpone medical care. An app that pretends to predict puts those users at meaningful risk.

Medical, legal, and financial claims. Adjacent to overpromising prediction, but worth naming separately. An AI palm reading app that suggests a health diagnosis, a legal outcome, or a financial event is making the kind of claim that real harm follows from. A user who interprets a “broken health line” reading as a medical warning may delay seeing a doctor. A user who acts on a “lucky in business” reading may make a financial decision they shouldn’t. The traditions of palmistry have never made these claims; the apps that do are inventing harms that did not exist before.

Retention and sharing of biometric-adjacent data. Palm reading photos sit in an ambiguous legal category — see our piece on biometric privacy in a palm reading app for the longer treatment. The ethical position is clearer than the legal one: keep photos as briefly as possible, prefer on-device processing, do not share with third parties without explicit consent. Apps that upload photos to remote servers, retain them indefinitely, or share them with advertising networks are failing a basic standard of user respect.

Manipulation of vulnerable users. Fortune-telling has always attracted users who are at vulnerable moments — facing illness, grief, financial pressure, relationship crises. The category invites this audience by design. The ethical baseline is: do not exploit it. Pricing tactics built on urgency or scarcity (“see your fortune NOW for $49.99 — offer expires in 5:00”), readings that fabricate negative outcomes to drive paid follow-ups, manipulative push notifications timed to user low-points — all of these are common in the category and all of them are abuses of the vulnerability the category attracts.

These four risks are not theoretical. They are visible in the current app stores. The category is being shaped by which builders avoid them and which builders profit from them.

What responsible AI fortune-telling looks like

The positive version of the same framework — what responsible practice in the category actually looks like — is a short list.

Symbolic, not predictive. Every reading, throughout, frames the practice as symbolic. The honest framing is that palmistry (or astrology, or tarot) describes patterns of temperament, not events. This framing has to run through the body of every reading — the language, the structure, the disclosure. It cannot be confined to legal fine print.

No medical, legal, or financial claims. Period. The reading can describe a temperament that relates to health, work, money, or legal decisions — but never makes claims about specific outcomes in those domains. If the user wants medical, legal, or financial advice, the app says so directly: get it from someone qualified to give it.

Clear data handling. On-device processing wherever possible. Minimum retention. Explicit consent for any sharing. A privacy policy that is readable in plain English, not legalese.

Honest monetisation. No urgency tactics. No scarcity tactics. No “this reading is special, but the full version requires payment” patterns. Pricing is transparent and what is paid for is the work, not the anxiety.

Care for vulnerable users. Recognition that the category attracts vulnerable users at vulnerable moments, and design choices that account for this. Crisis-adjacent readings are handled with extra care. Push notifications are conservative. The reading is calibrated to surface reflection, not to manufacture distress.

These five principles are not aspirational. They are achievable, and the apps that achieve them set the floor for the category. The apps that don’t are visible too — and the gap is the substance of the ethical question.

Why “entertainment only” disclaimers aren’t enough on their own

Almost every app in the category has a disclaimer somewhere — usually at the bottom of the screen, often in fine print, sometimes only in the terms of service — saying that the readings are “for entertainment purposes only.” This is the legal minimum, and it is far from enough on its own.

The reason is straightforward: a disclaimer in fine print does not undo a reading that has been written, throughout, in predictive language. If the app’s interpretive copy says “your future will hold,” “you will meet,” “a great change is coming,” “you will be wealthy,” the disclaimer at the bottom of the screen is contradicted by the body of the experience. The user does not read the disclaimer in a way that overrides the reading; they read the reading in a way that ignores the disclaimer.

Honest framing has to run through the reading itself, not just appear in legal text at the end. Specifically:

• The interpretive language of each reading uses symbolic verbs (describes, suggests, indicates, traditionally read as) rather than predictive ones (will, foretells, predicts).
• The structure of each reading offers a description of temperament, not a forecast of events. The reader is being shown a mirror, not given a fortune.
• The product UI signals “this is a reflection tool” through tone, pacing, and feature design — not just through a footer disclaimer.
• The upsells and follow-ups continue the symbolic framing rather than dropping it to drive paid conversions. “Get your detailed financial forecast” is a disclosure violation regardless of what the legal footer says.

PalmAura’s editorial discipline begins from this: a disclaimer is a floor, not a ceiling. The honest reading is the whole experience, not the legal text at the bottom of it.

For the broader frame on what AI palm reading can and can’t claim, see are AI palm readings accurate.

How PalmAura thinks about it

The five principles above are not policy decisions PalmAura considered and adopted. They are the structural design of the product.

The decision to frame readings as “symbolic, not predictive” came before the first line of code. The decision to treat photo handling as a privacy design problem came from the privacy framework, not from feature parity with competitors. The decision to disclose medical and financial limits in plain language came from reading the harm history of the category and refusing to participate in it.

This is not virtue signalling. It is a structural bet: that the long-term audience for AI fortune-telling will increasingly demand the responsible version, and that the apps that are built around responsibility from the start will be the ones that last. The apps built around aggressive monetisation of vulnerable users may produce short-term revenue. They are unlikely to produce a category that survives its current regulatory and reputational pressure.

PalmAura’s positioning — “symbolic, not predictive,” “private by design,” readings as “symbolic entertainment and self-reflection only” — is the operationalisation of the ethics. The aesthetic (Victorian-rooted, careful, restrained) reinforces the same message. The product is the ethics. They are not separable.

For the broader question of how AI is changing the practice of palmistry, see how AI reads a palm. For the historical context — palmistry as a 5,000-year tradition the current AI moment is one chapter of — see a short history of chiromancy.

An invitation to other builders

This piece is also an invitation, addressed plainly to other builders in the AI fortune-telling category.

The five principles named here are not competitive advantages we are guarding. They are the floor we think the category should adopt collectively, and adopting them is good for the category as well as for individual users. Apps that hold to these principles raise the average; apps that don’t drag it down for everyone.

The category is at a formative moment. The structural drivers — AI maturity, the Gen Z return to symbolic practices, the broader cultural appetite for reflective frameworks — are stable. The next few years will determine whether AI fortune-telling becomes a respected adjacent-to-therapy practice or a predatory app-store category that adult audiences learn to distrust. The choice belongs to the builders.

PalmAura’s bet is on the responsible version. We would welcome more company.

A hand has been read for 5,000 years. The technology has changed. The ethics have not.