A lot has changed since generative AI arrived in the world of photography in the autumn of 2022. According to surveys conducted over the last three years by the Association of Photographers, members reporting having lost work to generative AI had risen to 30% by September 2024. And by February of this year, it had leapt up to 58.1%. They know this, explains CEO Isabelle Doran, because of the openness of clients about their choice to save budgets by opting to use AI.

“It’s not that they’ve fully lost their jobs as such,” says Doran, “but it’s been reduced significantly to a point that it is very challenging for professional photographers to find commissioned work. But not just them, of course, everybody in that process; so models, retouchers, lighting assistants, studios, everybody in our ecosystem is being affected by this.”

While anecdotally, beyond the survey, the AOP are hearing about how much broader the impact is across the wider creative industries, there is a challenge in quantifying this. “We work in a freelance environment,” says Doran, “and so therefore what concerns me most is the fact that the government really doesn’t have a clear overview on what’s happening in our sector.”

I liken it to a kid in a sweet shop. You’ve got all these enticing new sweets that you can pick at and delve into, but as we know, they’re not necessarily good for your teeth

And without that clear picture, she worries, the government is unable to make informed decisions about the economic impact of generative AI on creators and those that work with them. This issue, Doran explains, is that the rapid adoption of the new technology, at the behest of big tech, is happening at such a pace that it’s reshaping the photographic and broader creative industries in a way that many in it – not least the government – do not fully comprehend.

“I liken it to a kid in a sweet shop. You’ve got all these enticing new sweets that you can pick at and delve into, but as we know, they’re not necessarily good for your teeth. Although that simplifies it, I think that’s the way that everybody initially looked at [generative AI] and thought ‘Oh wow, it can do this!’ when looking at the outputs generated by the machine, but not necessarily fully understanding what the machine has actually done.”

Doran argues that it is a misnomer propagated by big tech to anthropomorphise what machines do – that they take lots of different influences from the world in the way humans do. “When in fact generative AI programs are statistical pattern-predicting machines that are merely exploiting expressive works by millions of creators and performers without any of the ‘human’ creative effort and skill required.”

The AOP has been paying close attention to what exactly is happening in the gen-AI process, and how it is impacting photographers and those they work with. Vast swathes of creative works have been scraped off the internet to assemble datasets (such as the 500k+ datasets found on the open source site Hugging Face), which are being used to develop competing services.

Understanding what’s happening has been helped along, says Doran, in part by the transparency applied to one of the initial open source datasets, the Laion 5B dataset, a collection of captions and links to 2.3 billion images that were used to train Stable Diffusion and Midjourney. She cites later reporting that Midjourney further integrated the ‘styles’ of some 16,000 artists into training the program to enhance their service to users, and additionally enable users to upload someone else’s body of work.

“When you take that significant amount of input – which is everybody’s creative work, whoever’s uploaded work onto the internet – that’s all being consumed within these machines as data structures, it’s not the same as the way humans learn at all…. You’ve used so much labour – people’s labour – to create what appears to be this extraordinary tool, but it’s not a tool, it’s a competing service that everybody’s been heavily enticed to use. And it is now having a direct impact on commissioned work for commercial photographers and those who work alongside them.”

AI programs are merely exploiting expressive works by millions of creators and performers without any of the ‘human’ creative effort and skill required

Doran, and many photographers the AOP represent, consider this an injustice. “I think it’s fair to say that we’ve never seen a situation in which a sector is being so impacted by the fact that you’ve got an entity that doesn’t respect the investment and the hard work that creators, and the creative community, put into those works that are then published online and shared online and end up competing in the same online marketplace.”

She likens it to the food production industry: “You couldn’t imagine somebody coming along and just taking all the strawberries a farmer has painstakingly grown in a polytunnel without permission and then making their own strawberry milkshake, or giving the strawberries to somebody else who then makes their version of a strawberry milkshake, without any legal punishment for doing that. I mean, to the rest of us that’s theft, plain and simple.”

The issue isn’t necessarily copyright law – the United Kingdom is widely respected for what many refer to as the ‘gold standard’ in copyright law – it comes in its application in a new and quickly evolving landscape. Enforcement of the use of creators’ copyright-protected works is challenging – as we’re witnessing in the Getty Images case, the only UK case focused on generative AI to date.

It’s a competing service everybody’s been heavily enticed to use. And it is now having a direct impact on commissioned work for commercial photographers

While the UK government launched its Creative Industries Sector Plan – which aims to significantly increase investment by 2035 – the AOP and a coalition of others in the creative sector are campaigning to ensure individual creatives are themselves able to survive.

Recognising that “big tech has enormous resources to be able to lobby and persuade, all the way up to the top level of government, about why they need to be given an exception”, Doran saw the value in organising a cross sector alliance, along with a cohort of trade and union representatives.  This Creative Rights in AI Coalition (CRAIC), now numbering around 80 organisations from news media, design, fine art, music and broadcast, pools expertise and influence to ensure creators aren’t sidelined by the rush to automate.

As concern over gen AI’s unchecked use of copyrighted works grew, the AOP, alongside coalition representatives, supported and utilised a public facing drive to raise awareness of what they describe as the unlawful and unfair exploitation of creators’ works among key policy makers and the general public. Dubbed the Make It Fair campaign, its message was distilled into a simple call to action: “It is about trying to make it fair for those working in the creative industries to control, and have transparency, over the use of their works,” says Doran.  

By urging photographers, retouchers, designers, and other creative industry professionals to write directly to their MPs, the campaign has begun to turn individual frustrations into a collective voice. Beyond coalition building and grassroots mobilisation, CRAIC has pressed for formal commitments from Westminster. Together with key parliamentarians, they have secured pledges for economic and technology impact reports and have been invited to roundtable meetings alongside tech representatives.

Meanwhile, supporters in parliament represent a cross-party interest on this crucial issue, and include creators’ champion Baroness Kidron (cross-bench), together with a majority of House of Lords representatives, MPs Lee Barron (Labour), James Frith (Labour), Samantha Niblett (Labour), Polly Billington (Labour), Victoria Collins (Lib Dem), and Sir John Whittingdale (Conservative), to name a few.

The government has committed to convene three roundtable meetings – each bringing together a core agenda but different mixes of tech sector and creative industry representatives – to drill into AI training, transparency, data use and access, and inform next steps on the economic and technical reviews.

The first two of these discussions took place in July, and the remaining session will follow in September, feeding directly into the forthcoming AI and copyright consultation review and helping to shape a new fair framework for rights-holders, including photographers.

They’ve reversed their approach and decided to both reinvest in people and reinvest in shooting their clothing brand with human beings in mind

There are, Doran points out, some chinks of light that indicate that following an initial rush to drive efficiencies in image production – particularly in ecommerce – gen AI imagery might actually prove more costly for businesses.

One brand Doran is aware of, after initially deciding “let’s just embrace it” to chase cost savings, quickly discovered that post production costs and customer dissatisfaction with AI models outweighed any gains. “They’ve reversed their approach and decided to both reinvest in people and reinvest in shooting their clothing brand with human beings in mind.”

Beyond outright reversals, there are signs of a more considered, hybrid future –and even a framework for putting human creativity back at the centre of the industry. Bodies like the AOP are now “pushing for an ethical approach and trying to suggest a pathway that’s a bit more like the food standards process, where transparency is essential, so that you know the provenance, value exchange and the journey a product has taken to reach the consumer”.

This push for transparency, provenance and selective AI use offers a small but meaningful glimmer of hope for commercial photographers as the debate on AI rages on.

the-aop.org