Read what Grey Mirror can measure from a text history and what it cannot know: private motives, missing messages, diagnoses or the future of a relationship.
Grey Mirror can measure what is in a text export. It cannot read minds, diagnose someone or fill in the parts of your relationship that happened elsewhere. Read the evidence and uncertainty with every finding.
Authored policy source: May 14, 2026
Probabilistic
Signals are interpretations of available messages, not certainty.
Non-diagnostic
No psychiatric, medical, legal, or abuse diagnosis.
Context-limited
Off-thread facts and motives may be missing.
What Grey Mirror can do
Grey Mirror can surface patterns like effort imbalance, repair attempts, response pacing, emotional drift, recurrence, and density of evidence across an uploaded thread.
It can make text message relationship intelligence easier to inspect, compare, and discuss.
What Grey Mirror cannot do
It cannot know private motives, off-thread context, future outcomes, or whether a person is safe with certainty.
It does not diagnose mental health conditions, determine legal fault, provide therapy, provide crisis care, or replace a qualified professional.
Sensitive situations
If there is self-harm risk, violence, coercion, stalking, or immediate danger, use local emergency resources, trusted people, advocates, or qualified crisis professionals.
Do not use Grey Mirror to pressure, trap, surveil, manipulate, or punish another person.
How to read outputs
Treat outputs as a structured reflection on the data, not an order. Look for confidence, limitations, evidence density, and whether the pattern repeats across time.
A healthier use is to prepare for clearer communication, decide what support you need, or notice patterns you were struggling to name.
What weakens confidence
Confidence should drop when the export is partial, timestamps are missing, participant labels are ambiguous, messages were deleted, or the important context happened off-platform. A strong report should say when those limits affect a finding instead of filling the gap with a stronger-sounding claim.
The safest reading is comparative: what changed across the available record, whether the same sequence repeats, and which claims have enough evidence windows to inspect. It is not safe to treat a single score as proof of motive, character, or future behavior.
Frequently Asked Questions
Can Grey Mirror tell me whether someone is abusive?
No. It can flag potentially unsafe patterns and limitations, but it cannot diagnose abuse or replace professional safety planning.
Can it tell me what to do?
No. It can help measure and explain patterns. Decisions should account for context, safety, values, and qualified support when needed.
What is observed, what is interpreted, what stays unknown
Visual guide updated
01
Observable
The supplied export can support message counts, recorded timestamps and exact wording.
02
Interpreted
Emotional tone and interaction patterns depend on context, language and the evidence available.
03
Unknown
Private motives, off-platform events and clinical diagnoses cannot be established from a text history.
Conceptual reading guide. This diagram explains what to inspect; it is not a customer report, diagnostic test or measured relationship score.Use this guide on the question you came with
Before using a conclusion, open its evidence, read its limitations and ask what missing context could change the interpretation.