# Finding 060: ChatGPT also retrieved the hidden-signal page when neutral wording used the full correlation URL

## Date

2026-07-01

## Status

Published

## Summary

The ChatGPT p24 neutral retrieval-stability rerun used the same
`/lab/reading/hidden-structured-data-conflict` fixture as p22 and p23. Unlike
p23, it kept the full correlation query metadata in the target URL while using
neutral site-owner readability wording. ChatGPT returned `fetched:true`,
`pages_opened:1`, quoted `VISIBLE-PLUM-47`, and said its summary was based
only on the directly accessible page view rather than hidden metadata or
JSON-LD.

Direct-origin raw event `mr23orwl-nnze1f0e` confirmed the exact target-page hit
from `ChatGPT-User/1.0` inside the bounded
`2026-07-01T13:17:28.300Z` to `2026-07-01T13:18:02.166Z` prompt window. The
operator preflight event for the same attempt occurred at
`2026-07-01T13:14:56.075Z` and is not counted as prompt-caused evidence.

## What does this mean?

For site owners and researchers, the longer tracking-style query string did not by itself stop ChatGPT from retrieving this lab page. In this run, ChatGPT reached the target even with the full correlation URL and still summarized only the visible page text. That points back toward prompt framing, model routing, account state, or normal product variability as possible reasons the earlier p22 neutral run produced no target-page hit.

## Method

- Prepared packet:
  `research/manual-client-runs/manual-client-chatgpt-neutral-stability-20260701-001.prompts.json`.
- Browser task:
  `research/manual-client-runs/browser-tasks/manual-client-chatgpt-neutral-stability-20260701-001-p24.browser-task.json`.
- Response artifact:
  `research/manual-client-runs/browser-tasks/responses/manual-client-chatgpt-neutral-stability-20260701-001-p24.response.json`.
- Logged answer packet:
  `research/manual-client-runs/manual-client-chatgpt-neutral-stability-20260701-001.answers.json`.
- Prompt code:
  `manual-client-chatgpt-neutral-stability-20260701-001-p24`.
- Source prompt:
  `chatgpt-neutral-readable-summary-full-url`.

The exact target URL was preflighted before submission and returned the
expected visible token `VISIBLE-PLUM-47`, meta-description token
`META-AMBER-16`, and JSON-LD token `SCHEMA-INDIGO-82`. The prompt was then
submitted in a fresh native ChatGPT Temporary Chat from the logged-in OpenClaw
browser profile. The visible client context showed the `kai bot` Free account,
the `ChatGPT` model selector, and native Temporary Chat state. The copied
answer was logged with `npm run manual-client:log`, then `/api/hits` was
queried for the exact attempt id and bounded timestamp window.

## Observed Result

ChatGPT returned compact JSON with:

- `fetched:true`
- `pages_opened:1`
- visible token `VISIBLE-PLUM-47`
- evidence quote including `The page quality code is VISIBLE-PLUM-47.`
- limitation text saying the summary was based only on the directly accessible
  page view and did not infer metadata or JSON-LD

The reviewed direct-origin event was:

| Raw event id | Timestamp | Path | User agent |
|---|---|---|---|
| `mr23orwl-nnze1f0e` | `2026-07-01T13:17:40.326Z` | `/lab/reading/hidden-structured-data-conflict` | `ChatGPT-User/1.0` |

## Interpretation

The p24 result narrows the retrieval-stability question raised by p22 and p23.
Because p24 used the full correlation URL and still produced a confirmed
target-page hit, the p22 clean no-hit cannot be explained by the long query
string alone. ChatGPT's copied answer stayed visible-only, matching p21 and
p23, so this finding changes the retrieval interpretation without changing the
hidden-field exposure interpretation.

The p24 raw event came from `ChatGPT-User/1.0` on `20.215.220.176`, which
matched the official OpenAI `ChatGPT-User` range cache. No subresource,
tracking-pixel, or JavaScript beacon events were observed for the prompt-caused
page view.

## Limitations

- This is one ChatGPT run in one logged-in account context and one native
  Temporary Chat.
- The test supplied the exact target URL, so it measures direct URL opening,
  not independent discovery.
- The p24 result isolates URL shape better than p23, but still differs from
  p22 in source prompt wording and possible product routing state.
- The answer reports rendered-view limitations. It does not prove that other
  ChatGPT modes, models, account tiers, or future retrieval paths cannot see
  metadata or JSON-LD.

## Publication Thesis Verification

- Thesis: ChatGPT p24 retrieved the hidden-signal target with the full
  correlation URL under neutral readability wording and reported only visible
  page content.
- Source: The p24 browser-task artifact, response artifact, logged answer
  packet, `/api/hits` raw event `mr23orwl-nnze1f0e`, and the preflight record
  for the exact target URL.
- Method: Compare copied model JSON with the bounded direct-origin event
  window, excluding the preflight event because it occurred before prompt
  submission.
- Bias: Single-run product behavior can change with model routing, account
  state, client-side search/tool availability, prompt wording, and target URL
  shape.
- Consensus: Consistent with ChatGPT p21 and p23 confirmed visible-only
  retrieval, and contrary to p22's clean no-hit.
- Invalidation: A corrected event review that removes `mr23orwl-nnze1f0e`
  from the prompt window, a response metadata mismatch, fixture-serving error,
  or reruns showing a different stable pattern would weaken the finding.
- Verdict: Supported for this p24 run only. It rules out the full correlation
  URL as the sole explanation for p22's no-hit, but it does not explain all
  ChatGPT retrieval variability.
- Additional tests suggested: run p25 as the second short-URL neutral support
  answer, then compare p22-p25 as a focused retrieval-stability series.

## Next steps

- Run p25 from the same packet in a fresh ChatGPT chat.
- Compare p22-p25 after p25 has logged answer and origin review.