ChatGPT “Read” Coffee Grounds and Allegedly Revealed an Affair: What Does the Greek Divorce Story Prove?
At the end of June 2025, a story from Greece circulated in the media: a woman reportedly asked ChatGPT to interpret coffee grounds, and the answer linked her husband to another woman. The marriage later ended in divorce. Headlines can make it sound as if “AI uncovered an affair.” Most coverage repeated the husband’s account; no chat record or full relationship history was made available for independent review.
The more useful question is not whether coffee-ground reading can expose a secret. It is how a user should judge what an AI actually knows when it answers a question about an intimate relationship with confidence.
A circulating story is not a case file
Religion News Service’s June 30 report on AI and tarot mentioned the Greek story. The English-language Kathimerini report said the account came mainly from the husband’s public telling. The available information does not prove that ChatGPT uncovered an actual affair, or independently establish every reason for the divorce.
“ChatGPT said he cheated” should therefore be treated as a retelling of a personal story, not evidence that AI discovered a fact. A media anecdote cannot show that people commonly divorce because of AI readings.
A language model sees an image and a prompt
Coffee-ground reading is a tradition of symbolic interpretation. Given a photograph of a cup and an instruction, a model can identify visual features and generate an explanation in the style of a reading. The image cannot reveal another person’s private behavior. Producing a specific story does not turn it into an investigation.
When the question is whether a partner is faithful, the user may already be worried. If the model completes the story in the direction suggested by the prompt, that worry can feel externally confirmed. This is a limitation of generated answers and a risk in emotionally charged decisions; it is not unique to one divination tradition.
A picture should not decide a relationship
Doubts in an intimate relationship call for real communication, reliable evidence and the judgment of the people involved. AI may help someone organize questions or draft a non-accusatory way to start a conversation. It should not claim to confirm infidelity from astrology, tarot or coffee grounds.
Mystilink’s AI Agent can be framed as support for interpretation and reflection: helping a user separate observed facts, personal concerns and symbolic readings. Product capabilities should be described as they exist today; the system must not promise to diagnose a partner’s behavior.
A personal tone can make an answer more persuasive
People may read a fluent, direct response as the result of careful analysis. If a model writes as though it knows the people involved, users can forget that it did not investigate their relationship, ask the other partner or verify the facts. Products should show which inputs informed the response and avoid turning a guess into a conclusion on sensitive subjects.
Restraint does not require a useless answer. A system can suggest what might need confirmation, encourage the user to look for real-world evidence, help them communicate when it is safe, or point them toward professional support.
AI can help clarify a question, not establish a fact
The coffee-ground story spread in part because its headline combined an old divination practice, contemporary AI and marital conflict. The public reporting does not establish what, if anything, the AI uncovered.
For Mystilink, the brand value lies in tools people can understand and interactions that respect their autonomy. AI can take part in a symbolic conversation while making clear that it cannot investigate someone else’s private life.