Bug report 09 · fullscreen readable evidence images

If the Image Contains the Argument, the Image Must Be Click-to-Read

The bug this page documents

The evidence pages include text-heavy images. Those images are not decoration; they are part of the argument. When they appear only as thumbnails, the public reader cannot read the words, inspect the context, or understand what the document is talking about.

Actionable OpenAI product bug: when ChatGPT generates public HTML evidence pages with text-heavy images, it should automatically make every image click-to-read at full size. The user should not need to explain that readable images must be readable.

What the uploaded transcript already shows

The earlier transcript includes image-gallery patterns, thumbnail links, and lightbox-style CSS concepts for full-size image viewing. It also repeatedly frames the project as public documentation meant to be readable, summarizable, and shared. The product mistake is failing to carry that established image-reading requirement forward into the new bug-report pages.

Visual proof: this page is readable by design

v007 full-screen readable images contact sheet
Click this contact sheet. It opens in the same full-screen reader as every individual image below.

From the continuity-auditor perspective:

If a picture contains text, then the picture is text. If the picture is text, it must be readable. If the generated document makes it unreadable, the document has failed before the argument begins.

A public evidence page that traps text-heavy images in small boxes is not “visual design.” It is accidental censorship by thumbnail.

Required behavior going forward

New local image set for this report

Supreme Court termination rhetorical art
22. Supreme Court termination rhetorical art
images/22-supreme-court-termination-rhetorical-art.png
Consent, accountability, legitimacy triptych
23. Consent, accountability, legitimacy triptych
images/23-consent-collapse-accountability-triptych.png
New system for humanity: no debt, no fees, no control
24. New system for humanity: no debt, no fees, no control
images/24-new-system-for-humanity-no-debt-no-fees.png
Who controls the 4K and 8K screen
25. Who controls the 4K and 8K screen
images/25-who-controls-the-4k-and-8k-screen.jpg
War profit visual rhetoric
26. War profit visual rhetoric
images/26-war-profit-visual-rhetoric.png
Media trust and vulnerable audience risk
27. Media trust and vulnerable audience risk
images/27-grandma-media-trust-risk.png
AIPAC and Jubilee Coin visual rhetoric
28. AIPAC and Jubilee Coin visual rhetoric
images/28-aipac-jubilee-coin-visual-rhetoric.png
Schumer, AIPAC, and Jubilee Coin visual rhetoric
29. Schumer, AIPAC, and Jubilee Coin visual rhetoric
images/29-schumer-aipac-jubilee-coin-visual-rhetoric.png
Question everything / media literacy
30. Question everything / media literacy
images/30-question-everything-media-literacy.png
Future of news AI-powered TV
31. Future of news AI-powered TV
images/31-future-of-news-ai-powered-tv.png
System reset / public funding fairness
32. System reset / public funding fairness
images/32-system-reset-public-funding-fairness.png
Truth for the whole human race
33. Truth for the whole human race
images/33-truth-for-the-whole-human-race.png
AI control public-warning rhetoric
34. AI control public-warning rhetoric
images/34-ai-control-public-warning-rhetoric.png
Independent media and free press message
35. Independent media and free press message
images/35-independent-media-free-press-message.png
Voice of America AI interactive TV
36. Voice of America AI interactive TV
images/36-voice-of-america-ai-interactive-tv.png
Government branches fired rhetorical art
37. Government branches fired rhetorical art
images/37-fired-government-branches-rhetorical-art.jpg

2000-character bug report

OpenAI bug: public evidence pages containing text-heavy images must make those images click-to-read at full size. In this project, the images often carry the argument. A thumbnail gallery that cannot be expanded is not accessible documentation; it hides the evidence. ChatGPT should infer this automatically for public bug-report HTML and generated evidence packs. Add artifact rules: every image gets a local duplicate, readable filename, visible caption, click-to-fullscreen lightbox, open-image-in-new-tab link, Escape/outside-click close, and mobile-safe full-size viewing. Do not force the user to ask for this after the fact.

Evidence: [PUBLIC_URL_HERE]