# ChatGPT responded to a normal prompt as if it were a phishing email, malicious link and all — retained record

> Retained reddit record from reddit:ChatGPT.

Canonical: https://highsignal.app/data/records/9e09d40b053d7fa1

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ChatGPT)  published 2026-10-14 17:21 ingested 2026-10-15 02:37

[original source ↗](https://www.reddit.com/r/ChatGPT/comments/1x5x9ad/chatgpt_responded_to_a_normal_prompt_as_if_it/)

## retained text



```
Final edit: it was a well-hidden custom GPT called “Chat” that managed to be the top result on Google when I searched “xhatgpt”. Mystery solved. 

Never seen something like this happen, even scanned the link and it was verified malicious. Insane!

Edit: Report didn't land - "We have reviewed the content that you reported in ChatGPT and we found no policy violation. Thank you for your feedback."

Suggestions so far have been about custom GPTs or that my account's memory has been compromised. I looked through my chats, sessions, and usage history and nothing looks out of place there. All I can say is that nothing looks like unusual activity to my eyes and I haven't used any custom GPTs.

Edit 2: Investigated this even more, check this out:

# Investigation: Fake "OpenAI Security Team" phishing message in ChatGPT

**Bottom line: no local source was found that could have injected that fake message.** The machine is not compromised in any way that would alter the ChatGPT page. The text almost certainly came from the ChatGPT conversation itself.

# What I checked and what I found

**No local cause:**

* /etc/hosts — clean (only localhost). No DNS hijack.
* Proxy — none (env, gsettings, Chromium prefs all empty).
* LD\_PRELOAD / /etc/ld.so.preload — empty.
* No rogue processes, autoruns, cron, or shell-rc tampering.
* Browser history — the page was genuine https://chatgpt.com, not a lookalike.
* **TLS check:** openssl s\_client shows chatgpt.com presenting a real Google Trust Services WE1 certificate → **no TLS interception / MITM**.
* No cached network content contained "Security Verification Required" (grep of cache/storage found nothing).

**Extensions (Chromium is launched with** \--load-extension\*\*):\*\*

* Bitwarden — official, signature-verified.
* None match or modify chatgpt.com.

**No local reference to the phishing domain** anywhere in \~, the Chromium profile, /usr/share, /opt, /etc, or /var.

**The phishing domain itself:**

* gptverification.com was **created 2026-10-14 09:44 UTC — the same day, \~6 hours before the chat** — via NICENIC registrar, on Cloudflare NS. Classic disposable phishing infrastructure. It is not OpenAI.

**Corroborating history:** at 13:12–13:14 the same day, Google/urlscan/Reddit searches for gptverification.com were made.

# Conclusion

The evidence points to the phishing text being **generated server-side** — either a malicious/poisoned **custom GPT**, a shared-conversation prompt-injection, or a model output issue — not to anything on the computer. No local cleanup is needed.

# Recommended actions

* Do **not** visit the link; report the message to OpenAI (thumbs-down → report).
* Optionally disable Memory / Reference chat history for a truly clean context.
```



document fetched 2026-10-15 00:21 · canonical https://reddit.com/r/ChatGPT/comments/1x5x9ad/chatgpt_responded_to_a_normal_prompt_as_if_it
