AI, Identification, and the Long Game of Editorial Photography

There is a persistent myth that modern editorial photography has become effortless: point camera, press shutter, let artificial intelligence do the rest. The reality is rather different. AI is a tool, powerful, yes, but only when embedded inside a disciplined workflow built on verification, context, and experience.

I use paid versions of both ChatGPT and Claude. That matters, because the paid models remember preferences, tone, and rules. Over time you can train them to behave less like young, full of confidence, short on judgment news snappers and more like methodical investigators.. In my own workflow, AI is never given free rein. I supply the caption; it generates keywords. That division of labour is deliberate. Meaning comes from the photographer. Scale comes from the machine.

AI is also a sunk cost rather than a marginal expense. I use it across multiple parts of my professional and personal life, so its contribution to photography is incremental rather than something that needs to “pay for itself” on every image.

Alongside AI keywording, I use paid face-recognition software to help identify people entering and leaving government buildings Cabinet Office, Downing Street, Parliament. Face recognition is useful, but it is not authoritative. It narrows the search space; it does not end it. I always look for corroborating evidence: lanyards, security details, who else is in the group, and patterns of movement.

A recent example involved a visiting delegation to 10 Downing Street, from the Gates Foundation. I was initially confident about one individual but not enough to publish. When a second member of the same team was independently identified, the probability tipped into confidence. That second confirmation is critical. I never use a photograph unless I have at least two independent indicators of identity; unless, of course, the subject is already unmistakably (in)famous I don’t need to double check Amanda Holden or David Walliams

Research is where the real work begins. For both AI-generated keywords and face-recognition leads, I routinely check official websites, press releases, and, more timely, social media. Official reports are rare and usually appear days after the event, but social media frequently provides context in near real time.

I recently photographed a group of senior FCA officials without knowing the purpose of the meeting. A later social media post revealed it concerned the FCA’s use of AI. That single detail transformed the commercial and editorial value of the images. The same approach applies to diplomatic visitors. Number plates on ambassadorial cars provide the starting point; embassy websites or official feeds usually confirm the visit soon after.

This process is labour-intensive; pardon the pun, but it pays dividends in long-tail stock usage. I recently spent around three hours identifying Labour members of the House of Lords visiting 10 Downing Street. One image has already licensed. Others will, eventually. In stock photography, patience is not a virtue; it is a requirement. Some images wait for relevance. Some wait for history some just fade away.

There is also a collective intelligence at work. My friend and fellow photographer Richard Lincoln can identify the number plates of most UK ministers’ cars. That knowledge alone often tells you what kind of meeting is taking place, a National Security Council, emergency briefing, COBRA, routine Cabinet business. The presence and configuration of Personal Protection Officers, both Metropolitan Police and private, further narrows the field. It also provides advance warning of imminent ministerial or prime ministerial movements.

The same principles apply beyond politics. Celebrities, too, travel in patterns. Cars and minders are often more informative than faces at a distance. There are tactics for identifying protection teams, but those are not for public discussion.

What all of this underlines is that AI has not replaced editorial judgment; it has amplified it. The real value lies in combining technology with experience, scepticism, and a refusal to publish unless the facts are solid. In an age of automated everything, accuracy remains stubbornly human. A judicial mix of old school research and new technology seasoned with a large helping of hard earned experience provides evidence and truth.

That, in the end, is where photography still earns its keep.