Five decisions · one sitting · about two minutes
0 of 5 markedDo these five words earn a place on your list?
Every piece in the library carries one word that says what kind of thing it is: a brand film, a press release, a key visual. You approved that list in August, and you browse those words in named groups like Brand & Identity or Documentation & Technical. The library has been writing five words you never approved. Each row asks whether one of them earns a place.
What I need from you
For three of these I have suggested an answer, so you only need to say whether you agree. On the first and the last I could not call it either way, so those two are yours to pick.
If you disagree, type what you would do instead. That is what we will go with. If you would rather not decide today, leave it open.
You can answer straight from the row. Open a row to see the pictures at full size and the numbers behind them. The button at the bottom copies your answers out, ready to paste back to me in chat.
No recommendation from me. Neither way is better. 47 of the 62 are the picture inside an ad, and the other 15 are charts out of research posts. A Key Visual is the picture that carries a campaign, and it is already on your list. If you want a tiebreaker: would you ever look for that ad picture by typing the word infographic?
Charts out of research posts

OpenAI
See the page it sits in, on openai.com

OpenAI
See the page it sits in, on openai.com
OpenAI
See the page it sits in, on openai.com
OpenAI
See the page it sits in, on openai.com
Showing 4 of the 15.
Pictures out of the ad library

OpenAI
See it in LinkedIn's ad library

OpenAI
See it in LinkedIn's ad library

OpenAI
See it in LinkedIn's ad library

OpenAI
See it in LinkedIn's ad library
Showing 4 of the 47.
What the library holds today
| Pieces carrying the word today | 62 |
|---|---|
| Of the 62, read from the LinkedIn ad library | 47 |
| Separate ads behind those 47 | 7 |
| Pictures from the single biggest of those ads | 22 |
| Of the 62, charts off web pages and feeds | 15 |
Worth knowing before you answer. Those 47 ad pictures came from only 7 separate ads, and 22 of them belong to a single ad. There are far fewer different pictures here than the number suggests.
The working
You approved your words one at a time in August, and infographic was not among them. The library has been writing it anyway. 62 pieces carry that one word today.
So I looked at the pieces. One is a chart out of a research post: four line graphs: lifetime-in-tournament across opponent population sizes. It reads like a figure in a paper. Most are not like that. 47 of the 62 came from the LinkedIn ad library, which is the public archive of ads a company has run. There, each one is the picture inside an ad.
Some of those ad pictures hold charts too. One set is titled ChatGPT usage and adoption patterns at work. That is exactly the difficulty: the word would have to cover both a figure in a paper and a designed slide inside an ad.
That gap is the decision. If the word stays, it has to mean both, and a word that means two things is hard to browse by. If the word goes, each piece is looked at again and comes back as one of the words you did approve, most often Key Visual.
One more fact. Every one of these pieces already says whether it is a picture or a video. The word infographic is not what tells you that. It only claims what kind of picture, and on the ad pieces that claim is doing no work.
Nothing else about these pieces changes. The brand named on them, the campaign they belong to and the kinds of work noted on them all stay as they are. On the research chart that work is data visualization; on the ad it is graphic design. Nothing else leans on this word.
Nothing moves today. Either answer runs later, with the rest of the re-labelling.
Keep it. You would go looking for these by name, and that is the test that matters. The other way is to call it a way of shooting rather than a kind of thing. This is the easiest of the five to settle.
What these look like

OpenAI
See it on TikTok

See it on YouTube

OpenAI
See it on TikTok

OpenAI
See it on TikTok
Showing 4 of the 25.
What the library holds today
| Pieces carrying the word today | 25 |
|---|---|
| Of the 25, on YouTube | 16 |
| Of the 25, whose own title says behind the scenes, behind the design or bloopers | 21 |
One of the 25 does not belong here: a Google Doodle post about Teacher Appreciation Day, which is not behind the scenes at all. It is the second picture above, because the newest come first. The rest are what they say they are.
The working
25 pieces carry this word. The first picture above is OpenAI going behind the scenes with shy kids, the studio that merged Sora footage with live action for their short film. 16 of the 25 are on YouTube, so this is not one feed's habit.
Brands keep making these on purpose. 21 of the 25 say behind the scenes, behind the design or bloopers in their own title, and Nike alone has a long Behind the Design series.
The only real question is whether this is a kind of piece or a way of shooting one. A behind-the-scenes cut of a campaign film is, arguably, still that campaign film.
If it stays a kind, these 25 pieces keep their word and it joins one of your groups. If it becomes a way of shooting, the word comes off each piece and the piece goes back to being the film it documents. The footage, the brand and the picture do not change either way, and nothing moves today.
Add the word. It is what the industry calls this document. Publishing a safety document beside a new AI model is a different act from writing a white paper. The other way is to fold these into white paper and let the word go.
What these are
See where it came from, on openai.com
See where it came from, on openai.com
See where it came from, on openai.com
See where it came from, on openai.com
Showing 4 of the 14.
What the library holds today
| Pieces carrying the word today | 14 |
|---|---|
| Of the 14, read from the company's own release pages | 13 |
| Of the 14, read from a separate safety site | 1 |
| Of the 14, with any picture at all | 0 |
Not one of the 14 has a picture. Every one is a page of writing on a website, so this is the one row you have to read rather than look at.
What you have said about this before
“A model card, a technical report, a white paper, they are clearly output and part of something that we want to be able to capture and tag and analyze. They're not brand marketing creative verbal visual craft content.”— your review dictation of 4 August, asked whether a research paper is content
The working
14 pieces carry this word and they are the real thing. One is the GPT-5.1 Instant and GPT-5.1 Thinking System Card Addendum. 13 of the 14 come from the company's own release pages, and 1 from a separate safety site.
Until you named a Documentation & Technical group there was nowhere for a document like this to sit. That is why the word got invented. That group now holds white paper, specification, release notes and their neighbours.
So the question is narrow. Does that group take one more word, or is a system card a white paper with a different cover?
As a word of its own, the 14 pieces keep it and your technical group gains an entry. Folded into white paper, the word goes and these pieces are found under that instead. The document, its brand and its link stay exactly as they are, and nothing moves today.
Let it go. Vlog describes a way of shooting, not a kind of thing, and the pieces underneath are episodes of a brand series. It is close, though. The strongest argument for keeping it is one you made yourself, and it is inside the row.
What these look like

Nike
See it on YouTube

OpenAI
See it in Meta's ad library

OpenAI
See it in Meta's ad library

OpenAI
See it on YouTube
Showing 4 of the 9.
What the library holds today
| Pieces carrying the word today | 9 |
|---|---|
| Of the 9, on YouTube | 7 |
| Of the 9, running as paid ads | 2 |
| Of the 9, Nike Playlist episodes | 6 |
| Of the 9, OpenAI clips in the same style | 3 |
I told you before that these were all Nike Playlist episodes. 6 of the 9 are. The other 3 are OpenAI clips shot the same way, one a day-in-the-life video and two of them paid ads. The pictures above lean OpenAI, because the newest come first.
What you have said about this before
“it's a lot easier to check like, oh yeah, that got correctly classified as a vlog, as a interview, as a podcast, right?”— your March dictation on the review workflow
The working
9 pieces carry this word. 6 are Nike Playlist episodes, like Chipping with Golfer Lucy Li. 7 of the 9 are on YouTube and the other 2 run as paid ads.
Kind or look is a genuine coin flip here, and with this few pieces you can decide it on taste.
You said in March that it is easy to check whether something was correctly classified as a vlog. That is the strongest argument for keeping it, and it is yours rather than mine: a word you can check by eye earns its place even on a small number of pieces. I still lean the other way, because these are all episodes of a brand series, and that is the word I would look for.
Kept, the word stays on 9 pieces. Dropped, each one is looked at again and comes back as the series episode or brand film it is. The video, the athlete in it and the channel it ran on are untouched, and nothing moves today.
No strong call from me. 4 of the 5 are photographs from one developer conference, and the last is an ad the library filed here by mistake. Nothing moves today either way. The only cost of waiting is that more ads get filed under this word.
What these look like

OpenAI
See the page it sits in, on studiodumbar.com

OpenAI
See the page it sits in, on studiodumbar.com

OpenAI
See the page it sits in, on studiodumbar.com

OpenAI
See the page it sits in, on studiodumbar.com
See it in LinkedIn's ad library
Showing 5 of the 5.
What the library holds today
| Pieces carrying the word today | 5 |
|---|---|
| Of the 5, photographs from one studio's portfolio | 4 |
| Of the 5, ads filed here by mistake | 1 |
| Of the 5, with a picture | 4 |
One of the 5 is not a photograph at all. It is a Codex ad that the library read as an event photo. A word with this few pieces behind it is easy to file things under by mistake.
The working
5 pieces carry this word. 4 are DevDay photographs from one studio's portfolio: the venue atrium with its ASCII flower mural, the keynote hall, the bracket wayfinding sculptures and the inflatable frog mascot.
The one that is left is a Codex ad, Take everyday work from idea to finished output with Codex, which the library read as an event photograph.
Your Events & Experiential group already holds Event Film, Keynote and the booth or installation piece. So the pick is whether photography joins them, or whether these are key visuals that happen to have been shot at an event.
As a word of its own, event photo joins your events group and the 5 pieces keep it. Dropped, the photographs become key visuals shot at an event. Either way the stray ad is looked at again and comes back as an ad, because it was never an event photo.
The photographs, the studio credit and the links to where they were published are untouched, and nothing moves today.
Every number and every picture here came out of the live library on 7 September 2026, and each picture was fetched and checked before it was shown. Looking at this page changes nothing, and neither does answering it today.