Should Your Viewbook Use an AI-Generated Photo?

By Midya U · Midya U Advisory

In brief: In September 2026, Stanford apologized after generative AI erased a student from his own photo on a welcome banner — even though the university already had a policy against exactly this. Every photo an institution publishes is a claim about something an applicant is paying for, and Canada's ad-industry code now expects AI-altered images to be disclosed. This piece covers where generated imagery is actually getting into institutional marketing and the one-page policy to write before your next shoot.

In September 2026, a Stanford student named Billy Ramirez was sent a photo by a friend. It was a photo of him and his friends taken back in 2024 while eating at a Lunar New Year dinner. Only he wasn't in it anymore. Generative AI had replaced him with someone of a different race and gender, slimmed the faces of the other two students and dressed all three in Stanford merchandise for a welcome banner in student housing.

"Seeing my identity changed and being left out of the picture made me feel, in a way, silenced and erased," he told The Stanford Daily, which reported the story on September 23. Stanford took the banner down and apologized. "Quite simply, this was wrong and should never have happened," a spokesperson told NBC News.

Stanford already had a rule against exactly this. Its AI policy says "the use of AI in producing or altering images of Stanford people, events, research, facilities, or achievements is strictly prohibited." This mistake came from Residential and Dining Enterprises, the office that runs student housing and dining, and the rule didn't stop it.

Why does one altered photo matter to every university?

Because every photo you publish is a claim about something the applicant is paying for.

An applicant can't try your program before they pay for it, so they judge it on what you show them: the residence room, the lab bench, twelve students around a seminar table, the campus in October light. If any of those images shows something that isn't there, you've made a claim you can't back up to someone who's about to spend tens of thousands of dollars on it.

Students are also checking. Keystone Education Group's 2026 survey of more than 67,000 prospective students found 72% trust university websites, against 30% for AI. I wrote about it in The Trust Gap AI Hasn't Closed: students increasingly hear about you from an AI assistant and come to your site to see whether it's true. A generated campus photo can cost you that trust.

Where is AI imagery getting into institutional marketing?

In more places than the marketing lead usually knows about.

• Clean-up edits. Construction fencing removed, grey skies swapped for blue, an empty lecture hall filled. Each one feels harmless, and after a few years of them the campus in your brochure is one nobody can recognize.
• Other offices. Housing, athletics, faculties and student services make their own banners, posters and social posts, often with their own tools. Stanford's banner came from one of them.
• Agency and stock work. A lot of "stock" is now generated, and plenty of agency contracts won't mention it.
• Ad platforms. Meta and Google can generate image variants of your ads automatically unless someone has switched that setting off.
• Creators and ambassadors. Axios reported on September 8, 2026, on AI-generated fitness influencers selling products with bodies that don't exist, and noted that platform labels on Meta and TikTok aren't foolproof.

Each of these usually begins as a shortcut someone took on a deadline or a way to deal with a budget cut.

What should an imagery policy say?

It should fit on one page and answer three questions: what is never generated, what can be, and what counts as an edit.

• Never generated or materially altered: real students, staff and faculty; your campus, buildings, residences and facilities; class sizes; graduates at work. If a reasonable applicant would rely on the picture to decide, it has to be a photograph of the real thing.
• Allowed, with a label where it's not obvious: abstract backgrounds, illustrations, icons and concept images that nobody would mistake for a photo of your campus.
• Edits: colour, cropping and exposure are fine. Removing or adding anything a visitor would see if they came in person is not.
• Provenance: who shot it, when, and consent on file for every person in frame, kept somewhere you can find it in two years.
• Everyone else follows it too. Put the policy into your agency contracts, creator agreements and ad-platform settings, and send it to every office on campus that makes its own materials.

Writing the policy is the easy half. Stanford's policy was already on the books. Its response, according to the university's public relations director, is training for all staff and "proper review of all materials," which suggests the gap was in who knew the rule and who checked the work before it went up.

Do you have to label AI-generated images in Canada?

Sometimes, under the industry code rather than law.

Canada has no federal AI law. The Artificial Intelligence and Data Act died when Parliament was prorogued in January 2025. What exists is self-regulation. Ad Standards updated its Influencer Marketing Disclosure Guidelines in October 2025 to say that if an image showing a product's benefits or attributes "is generated by AI, or significantly altered using AI, this should be disclosed," with #MadeWithAI or #AIcreated as examples. AI-generated influencers carry the same disclosure rules as human ones, and the guidelines apply to brands as well as influencers.

If you run a student ambassador program or pay creators, you're inside that guidance now. Whether it reaches a photo in a printed viewbook hasn't been tested, as far as I know, and I wouldn't build a policy that depends on the answer. The safer line is the one above: don't generate the things applicants rely on, and label the rest.

Where would I start?

With the images already live. Take your ten highest-traffic program pages and the current viewbook, and for every photo answer three questions: who took it, when, and what's been changed since. Anything nobody can answer gets replaced or labelled. Then ask the other offices on campus that print banners and posters to do the same with theirs.

Then write the one-page policy and send it to your agency before the next shoot. If your photography is carrying a brand story that no longer matches the campus, How to Tell If Your Brand Predates Your Market is the place to start, and it's the same ground I cover in Brand Repositioning.