What Does an AI System Know About You? Authenticating the Individual Creative's Unique Identity
Updated: 6 days ago
By Hart Getzen — executive portrait photographer, producer, and publisher. Austin-based, working nationwide. PPA member; work held in the collection of the New York Historical.
Every working creative's name lives on several platforms at once: a credits database, a personal site, professional directories, festival and trade coverage. Each holds something true; none was built to agree with the others. That is the normal condition of an active career. Authentication is the work of claiming that record as one identity, and it turns out to be the same claim a photograph makes when it carries its own provenance: this is a real person, this is who made it, this is verifiably who they are. This Insight works it through on one real, current example: an Austin writer-director with a decade and a half of work across film and commercial production.
What is an AI system building when it looks you up?
Search your own name. Then ask an AI assistant who you are. The two answers are rarely built from the same material, and neither is written by you. Both are assembled from whatever the machine can find, in whatever form it finds it, weighted by whichever source happens to be easiest to index. Neither one checked with you first, and neither feels obligated to be right.
For someone early in a single-company career, that material is usually thin and already aligned: a company bio, a LinkedIn profile, a press mention, all pointing one direction. For an individual creative — a director, a writer, an architect, a designer, a working artist — there is simply more of it, because there is more career. A credits database. Professional directories. Festival programs and trade coverage going back years. A personal site carrying the current, fullest account in the creative's own voice.
This is the normal condition of a prolific working career
It is worth saying plainly: a record spread across platforms that describe you slightly differently is not a sign that anything was neglected. It is what an active career produces.
Every platform was built for its own purpose and indexes what it was designed to index. A film database tracks film credits. A directory tracks employment. A festival program describes a single project at a single moment. A personal site carries the current, authored version. None of them was designed to reconcile with the others, and none asked permission before publishing. The busier the career, the more platforms hold a piece of it, which means a scattered record is, if anything, evidence of range.
What has changed is the audience. For most of the time these platforms have existed, a human being did the reconciling. A producer, a programmer, a journalist who read three sources and understood they described one person. Increasingly the first reader is a machine, and machines do not infer. They match. Authentication is how you give them something unambiguous to match against. A machine will happily treat a middle initial as grounds for splitting you into two people, and it will do this with total confidence and zero self-doubt.
Why I am writing this series
I came to this subject through my own record, not a client's.
Before founding Hart Getzen Photography, I spent twenty years as a senior producer and writer in motion picture and large-format media production: at IMAX Corporation, at Trumbull Studios with Douglas Trumbull, as a technical manager at Francis Coppola's Zoetrope Studios, and in development at 20th Century Fox. Producer, writer, technical manager, development producer, and now photographer and publisher. Several disciplines, one continuous career.
That range is the kind of thing a person is glad to have. It is also, as I found out, exactly the kind of history that machines struggle to read as one person. Each of those roles was indexed somewhere different, in that platform's own vocabulary, for that platform's own purpose. Nothing tied them together except me.
It took many months of work to authenticate my own record, deciding which profiles to claim, which version of my name and titles would be canonical, and how to state a career that crossed disciplines so that it read as range rather than as several different people who happened to share a name. There was no simple guide for it. There still is not much of one.
That gap is widening, because working across disciplines is no longer unusual. It is closer to the norm. Directors produce. Photographers publish. Artists design, teach, and consult. Writers direct. The single-discipline career that most databases were built to describe is becoming the exception, and the people with the most range are the ones the machines read least well.
So this series is the guide I wanted and could not find. Everything in it is something I worked out on my own record first, and the aim is to make it simpler for everyone who comes to it.
A record worth authenticating: Christian Rousseau
Christian Rousseau is an Austin writer-director whose career has run on two tracks at once for a decade and a half, and that is precisely what makes his record worth looking at.
He came out of the University of Texas at Austin with a degree in Radio-Television-Film. His senior thesis, The Overman, became a laureled professional debut and that same year he founded San Austin Productions. The commercial work since has carried an international client list: Ferrari, UPS, Facebook/Meta, Dole Fruit, and Ozarka among them. Concurrently, he's crafted his own films: a feature heist thriller NO LOSS // NO GAIN, which he wrote, directed, and produced; headed to festivals this year is a comedy short film, The Interventionists; and his sci-fi horror script Bleak Oblivion, which garnered several prizes at Emerging Screenwriters, ScreenCraft, and Big Apple Film Festival, is now seeking investors and financiers.
Two tracks, one filmmaker. That is a strong career by any measure. It is also a record no single platform was ever built to hold: a film database captures one track, a professional directory captures fragments of the other, and only his own site carries both. Authentication is what makes those pieces resolve to one person.
"Stories are vital to everyone. They’re in our DNA. I feel it’s my job to contextually apply that to every aspect of my work: how a shot tells a story; how a schedule serves the story; what story a character is telling themself. Whether narrative or commercial, that's the space I live in: the myths, legends, and truths we tell ourselves. Keeping that mindset is how I find the honest, earnest version of any project."
Authentication in practice
What would authenticating that record actually involve? Four moves, each addressing something the record already contains.
The name. Christian's credits-database profile carries the notation the platform uses when two professionals share a name exactly — a separator that keeps neither one's credits from being absorbed into the others. It is the database protecting his record, not flagging anything about it, and it is permanent by design. Authentication means claiming the profile so he governs what it says, and publishing a canonical bio on his own site with structured Person markup that names it, so the profile and the authored account are readable as one person. (A forthcoming edition in this series covers the mechanics of claiming, and what cannot be changed no matter how long you spend on it.) Think of it as the database's way of saying "not that one," and, once assigned, it holds that position forever, unmoved by argument, apology, or a strongly worded email.
The title. His feature carries authored brand identity on his own site, styled NO LOSS // NO GAIN, while the credits database lists it in standard industry format. Both are correct for their context. Authentication means designating one canonical styling and stating the relationship between the two explicitly, so a machine reads one film with an authored treatment rather than two loosely similar entries.
The full body of work. A film database is built to track film and television credits, which by design means it does not carry the commercial and corporate directing that makes up the other half of his career. That is the boundary of what such a database is for, not a gap in the work. Authentication means giving the complete working record an authored, verifiable home that search and AI systems can find and cite.
The source. Directories and aggregator profiles are engineered to be indexed, so they surface readily for any working professional's name. Authentication means making his own site and portraits the strongest, most current, most clearly attributed account available, because search and AI systems weight authority and coherence, and a verified canonical source outranks a compiled one on the merits.
What authentication looks like
Nothing here is generated or invented. Every fact that matters already exists somewhere in the record: a name, a title, a client list, a body of work. Authentication is the work of verifying it, stating it consistently, and making the authored version the one that leads:
● one canonical name and title styling, used identically across your new portraits, your personal site, and every bio and caption that runs with them
● a canonical bio page stating plainly and confidently who you are, distinct from any same-named professional, linking out to the profiles that are verifiably yours
● Person schema, the structured markup which search and AI systems read directly, tying your site and verified profiles to one confirmed entity
● descriptive filenames, alt text, and captions on every new portrait, naming you, your role, and your work the same way each time
● embedded metadata carrying the declaration used across this series: an original photograph, an original digital capture from real life, not AI-generated and not licensed for AI training.
Where the photograph and the record make the same claim
Those last two lines are not housekeeping. They are the reason a photographer is writing this at all.
A finished portrait is not only an image. It is a file, and a properly prepared file carries a set of industry-standard provenance fields written into it: who made it, when, where, under what rights, and, critically, a declaration that it is an original digital capture sampled from real life. That last field is a formal, machine-readable assertion that the photograph is a photograph: not generated, not synthesized, not assembled from a prompt. It has an established standard behind it, and it travels with the file wherever the file goes. Refreshingly, it is just a person who stood still while someone else pressed a shutter button.
Set that beside what the record work does. A canonical bio page with Person markup asserts: this is a real person, these profiles are confirmed as such, this is the account he authored. The provenance fields in the image assert: this is a real photograph of that person, made by a named photographer on a known date. Two assertions, one structure. Both are claims of verified authorship, made in a form a machine can read without inferring anything.
This is why the portraits and the record cannot sensibly be treated as separate projects. An authenticated record with generic or unattributed images is missing the part that shows a face. A beautiful portrait sitting in a file with no creator, no rights, no declaration of origin, and no connection to the person's authored account is a strong image doing none of the work it could do. Neither is complete alone.
And the timing matters. As more of what circulates online is synthetic, the ability to demonstrate that an image is an original capture of a real person stops being a technicality and starts being the point. The photograph that can prove what it is will be worth more than the one that cannot, not because it is better made, but because it can answer the question.
What will change, and what will not
A fair question to ask of any piece written about AI systems: how long is this good for?
The specifics will move, and quickly. Which platforms offer verification, what a subscription tier includes, which systems read which signals. Expect all of it to shift, more than once. If you enjoy memorizing platform policy pages, these next few years will be a delight. If not, keep reading. The part that matters is below.
The foundation will not, and that is the more important half. Nothing in the underlying work was invented for the AI era. The structured-data vocabulary that describes a person has been stable for over a decade. The metadata standard that carries authorship inside a photograph predates the commercial web. Stating your name and title the same way everywhere, publishing an authored account on a page you control, linking verified profiles in both directions, and describing every image accurately — none of that was devised for answer engines. It was good practice for being understood, and answer engines simply became the newest thing doing the understanding.
So build the durable layer first. A record built to be coherent does not need rebuilding each time a system changes, because every system is reading for the same thing: a clear, consistent, verifiable account of who someone is and what they have made.
Why this article fits in the series and why you do not need to read the other two to use it
The portrait system piece covers one person's core images. The (forthcoming) image library piece covers an organization's full record: people, place, process, tools. This is neither. It is a single, distinctive career, already visible across more platforms than most, being authenticated as one identity, and photographed so that the images carry the same claim the record does.
Frequently asked questions
What does authentication actually mean here?
Two things at once. For the record: claiming the profiles that are yours, designating canonical language for your name and titles, and publishing structured markup that resolves to one confirmed identity. For the photographs: industry-standard provenance metadata declaring each image an original capture of a real person by a named photographer. Both are verifiable claims of authorship.
How does this relate to AI-generated imagery?
It is the opposite of it and says so in machine-readable form. Every frame is an original photograph, and each file carries the standard declaration that it is an original capture from real life, plus rights terms prohibiting AI and machine-learning training while leaving the images fully legible to search and AI systems. AI-ready means AI-readable, never AI-generated. A distinction worth having, now that a fair share of what circulates online was dreamed up by something that has never left the building.
Does this apply only to filmmakers?
No, and it is not really a question of discipline. It is a question of how many systems have recorded your work.
The pattern holds for anyone whose career has been indexed in more than one place by platforms that never spoke to each other: directors, writers, visual artists, architects, designers, and musicians, certainly; but equally a founder on a second or third venture, an executive with patents, trade coverage, and conference credits behind them, or a manufacturer whose work is listed by industry associations, supplier directories, and standards bodies.
This piece uses one director's career because it is a concrete, current example, not because the situation is unusual.
What if my career crosses several disciplines?
Then this matters more, not less. Multi-disciplinary careers are the hardest for machines to read as one person, because each discipline tends to be indexed by a different platform in a different vocabulary. No single database has a dropdown menu wide enough for you, and that is a compliment. The work is the same; there is simply more to align, and more to gain from aligning it.
I am a founder or an executive, not a creative. Does this still apply?
Often more than expected. A first-time founder at one company usually has a clean, aligned record. But a second or third venture, an acquisition, a name change, a patent filed under a former employer, a decade of trade-press mentions, and a string of conference listings produce exactly the same fragmentation, with the added problem that some of the strongest-ranking sources are prior-company pages you no longer control. The work is identical: one canonical account, consistent naming, verified profiles linked in both directions, and original photographs that carry their own attribution.
What comes next in this series
This piece stays at the level of principle. Later editions get specific, and two are already scoped.
One covers the platforms directly: which listings can genuinely be claimed and how, which cannot be changed no matter how much time you spend. The automatic disambiguation notations, the permanent identifiers, the listings with no correction path at all and how to tell the difference before you lose a month to the wrong one.
The other is a worked Person schema block, written out in full and explained line by line, so it can be copied rather than commissioned.
Neither is a prerequisite. If you do only what this article describes, settle on one canonical version of your name, your titles, and your body of work; claim the profiles that are genuinely yours; publish the authored account on a page you control; and make sure every new photograph carries its own description and its own declaration of origin, you are already most of the way to a solid, powerful, AI-ready identity. The later pieces sharpen the edges. This one gets you standing on solid ground.
A photograph shows what you look like. An authenticated record, and an authenticated photograph, make certain that the identity attached to it is verifiably, unmistakably yours, recognized correctly by people and machines alike, on the first look.
Hart Getzen builds executive portrait systems, institutional image libraries, AI-ready image libraries, and printed brand editions for founders, leaders, artists, and institutions. To discuss a commission, get in touch.
Comments