It works, but it is not private. ChatGPT, Claude, Gemini and most other assistants accept file uploads, so they can read your Instagram export and list who does not follow you back. The cost is that a complete, dated map of your social circle now sits on an AI company's servers, often under settings that allow training unless you switch them off.
Can ChatGPT actually find who doesn't follow me back from my export?
This tip is all over social media now: download your Instagram data, drop the followers and following files into a chatbot, and ask "who doesn't follow me back?" If the assistant runs code on the files, the answer can be correct. If it just reads the text and replies, long lists are exactly where a language model can skip or invent a name, and you have no easy way to see which happened.
But whether it works is the smaller question. The bigger one is what you gave away to get the answer.
What's in my Instagram data export?
A followers and following export is not a count. Every entry holds a username, a link to the profile, and a timestamp for when the follow happened. With 3,000 followers that is 3,000 real people, plus the date each connection started. The breakdown is in what's inside your Instagram data download.
Plenty of people do not stop at that section. If you exported everything, the ZIP can also hold your direct messages, your posts, your comment history, and more. Drag the whole folder into a chat window and all of it goes along for the ride.
And most of that data is not only yours. Your followers never agreed to be uploaded to an AI service.
What happens to my file after I upload it to an AI?
Everything below comes from each company's own help pages as of this writing. Policies change, so check the current version before you rely on any of it.
- 1.Your upload can train future models. OpenAI's Data Controls FAQ says new conversations are kept out of training only once you turn off "Improve the model for everyone." Perplexity's data collection page says AI Data Retention is on by default for Free, Pro and Max users. On X, a setting under Privacy and safety, Data sharing and personalization, controls whether your Grok interactions are used for training.
- 2.A person may read it. Google's Gemini privacy notice says a subset of chats is reviewed by human reviewers, that reviewed chats are kept for up to three years, and asks you not to enter confidential information you would not want a reviewer to see.
- 3.It stays around. Microsoft's Copilot privacy FAQ says conversation activity is stored for 18 months by default, and that files you share are treated like any other conversation.
- 4.Where it is stored matters. DeepSeek's privacy policy says it collects, processes and stores personal data in the People's Republic of China.
- 5.Meta AI is a special case. Meta already holds your follower graph, so asking Meta AI about it is not new exposure. But Meta said that from December 16, 2025 it would use your AI chats to personalize content and ad recommendations in most regions, so what you ask it feeds your ads.
Claude handles this differently: Anthropic's privacy center says whether consumer chats are used to improve Claude depends on your privacy setting, and incognito chats are excluded. Either way, the file has still left your device.
Do ChatGPT, Claude and Gemini train on my uploads?
| Assistant | Training on your chats | Where to change it |
|---|---|---|
| ChatGPT | Yes, while "Improve the model for everyone" is on | Settings, Data Controls |
| Claude | Depends on your privacy setting | Settings, Privacy |
| Gemini | Some chats reviewed by people | Gemini Apps Activity |
| Microsoft Copilot | You can opt out | Copilot privacy settings |
| Perplexity | On by default | Settings, Preferences, AI Data Retention |
| Grok on X | Controlled by a data sharing setting | Privacy and safety, Data sharing and personalization |
| DeepSeek | Policy offers an opt-out right | See its privacy policy |
Turning training off helps. It does not change the fact that the file left your device.
How do I find unfollowers without uploading my data anywhere?
What you want is a comparison: the accounts you follow minus the accounts that follow you. That does not need a language model, a data center, or anybody's training pipeline. It needs a parser.
WhoUnfollowed reads your Instagram export inside your browser. The ZIP is never uploaded, and no AI model is involved at any step. You do not have to take our word for it, either. The parser is open source under the MPL-2.0 license and the code is on GitHub, so anyone can read exactly what happens to the file. With a chatbot you can read a privacy policy. You cannot read the system.
The same goes for tools that skip the export and ask for your Instagram password instead. Those are worse than a chatbot, not better.
In Short
- A chatbot can read your Instagram export and list non-followers.
- The file holds every follower with dates, and a full export can include your DMs.
- ChatGPT and Perplexity train on consumer chats by default unless you opt out. Gemini sends some chats to human reviewers. Copilot keeps history for 18 months by default.
- WhoUnfollowed does the same comparison locally, with open-source code and nothing uploaded.
You want a list of names, not a new copy of your data on someone else's server. Get it in your browser in about two minutes.
