Methodology v0.1

Evidence before impression. Always.

We review the public presentation of an Instagram profile. We do not judge the value, character, or safety of a person behind it.

The publication threshold

Every public conclusion must be explainable, sourced, dated, and confirmed by a human reviewer. “It looks AI” is a lead, not a finding.

01

Confirmation

Direct

A self-disclosure, creator statement, official portfolio, or other primary source explicitly identifies the persona as synthetic.

02

Independent signal

Strong

A significant, repeatable indicator such as matching creator provenance or material identity inconsistency across independent sources.

03

Context

Supporting

Relevant context that can reinforce a case but cannot determine a verdict on its own.

04

Not decisive

Weak

Aesthetic impressions, follower patterns, missing livestreams, or a single detector result never decide a public verdict.

Verdict rules

AI identity confirmed
Requires at least one direct, publicly verifiable source.
Likely AI-generated identity
Requires at least two independent strong sources, or one strong source with multiple independent supporting sources and explicit reviewer reasoning.
AI-assisted or mixed identity
Requires evidence of a real operator or subject and evidence of material AI transformation in how that identity is presented.
Inconclusive
Not published as an accusation. Insufficient evidence means no verdict.

Review history and corrections

Published reviews are not silently rewritten. A material correction creates a new review and the earlier version remains in the history. Evidence may be hidden when privacy, safety, or copyright concerns require it.