Keep the reached public profile and the supplied reply separate. From a willingly shared screenshot or your own existing permitted conversation, record the AI-response label actually shown for that reply. Keep the original line attributed to a generated response. If the human contact later confirms its content, record that as a separate statement they made; it does not establish that they personally typed the original reply or that its claims are true.
Find public social profile clues with our TraceSeek app.
Begin with the two things you actually have
Start with material already in hand: a public username, photo or profile link a dating contact chose to share, plus a screenshot willingly provided to you or your own existing permitted conversation showing an AI-response label. Open only the known public profile; do not access anyone else's account or request a partner's chat archive. Record the profile reached, then separately transcribe the label actually displayed for the supplied reply and note its visible context. These are observations about a profile and an artifact, not an invitation to retrieve messages.
Why a correct profile does not settle the speaker
Current platform material explains that creator AI answers common direct-message questions and story replies, and that those responses are labeled. That tells you a labeled generated response is a recognized mechanism inside a creator's Instagram experience, so the reply can sit under a real account without the account owner having typed the sentence. The profile match can be entirely right while the belief that the creator personally wrote a relationship claim remains unsupported. Record the generated statement as an AI response. Nothing about the label establishes who operated the account, why the message appeared, or whether any biographical detail in it is true. [Create Your Own Custom AI With AI Studio]
| observation | limit | next step |
|---|---|---|
| Shared username, photo or link opens a public Instagram profile | Reaching a profile does not verify who controls it or that the person you are dating is its owner | Note the profile and the public details visible there; keep it separate from the reply |
| A reply carries a visible creator-AI label | The label marks a generated response, not the creator's personal assertion | Copy the label exactly and treat the line as AI output, not a human statement |
| A screenshot shows the reply but the label is missing or cropped | No clear label means the speaker cannot be assigned from that material | Leave the line unattributed and end the check rather than guessing |
Apply the lifecycle limit without hunting for setup steps
Meta's August 10, 2026 update says new AI Studio characters can no longer be created and existing ones cannot be edited, while existing characters and chats remain active. This limits the older creator-AI description on the same page: its original creation and setup steps are not current instructions. Do not prompt an AI to test the reply. This source also does not establish universal current availability, exact label wording or a particular control in your app. [Create Your Own Custom AI With AI Studio]
Leave unclear source material unattributed
A cropped screenshot, a forwarded image, or a line with no visible label does not identify a speaker. Write it down as unattributed and stop. Equally, a visible label does not validate the rest of the screenshot, the surrounding messages, or anything the profile says about itself. Keep your three-part record clean: the public profile reached; the exact label as displayed; the statement that stays an AI response rather than an authenticated personal assertion by the profile owner. If that third field reads as AI output, do not quietly rewrite it later as something the person said.
Ask one person, then stop
If the distinction matters to a decision you are making, use a route the contact has already opened and is willing to use. Ask about one specific line and one specific question: which statements in that exchange did they personally write? Do not message the AI to test its fidelity and do not keep prompting it for private details. A hypothetical example: you might say you saw a labeled automated reply and want to know which parts of the conversation were theirs. Whether that conversation continues is their choice. If there is no clear label or no permitted source, leave attribution unknown and end the check here.
- Keep the public profile link, the verbatim label and the quoted reply in three separate fields
- Ask one question through an existing willing channel: which lines did you write yourself?
- If the label is missing or the image is cropped, mark the speaker unknown and stop
A couple more questions
The shared Instagram profile is clearly this person's. Does that make the labeled reply their own statement?
No. Reaching the right public profile and knowing who produced a reply are separate questions. A labeled creator-AI response is a generated answer shown through a creator's Instagram experience, so the profile can match while the sentence itself remains AI output rather than an authenticated personal claim.
The screenshot shows a reply but the AI label is cut off. Can I assume a person wrote it?
No. A missing label does not show that a human wrote the line, and a visible label elsewhere does not cover the cropped part. Record the reply as unattributed and end the check, or ask the contact directly which lines they wrote.
Should I create a creator AI or prompt one to see how it behaves?
No need, and creation is not a current option anyway: new AI Studio characters cannot be created and existing ones cannot be edited, though existing characters and chats remain active. Use only the labeled material you already have and keep prompting out of your verification.
What should go in my record, exactly?
Three fields: the public Instagram profile your shared input reached, the exact AI-response label as displayed, and the quoted statement marked as an AI response rather than a personal assertion. Anything unlabeled or cropped stays unattributed.
Sources & editorial notes
Official references checked on October 25, 2026. Checklists, comparison tables and example wording are DateClues editorial tools, not official tests or validated risk scores.
General information. No guide or tool can certify a person’s identity, intentions or safety. Read our editorial standards.