In the second half of 2022, when artists discovered that their work had been swept up and used to train models for AI image generators without their consent, it didn’t take long for activists to organise themselves against the practice.
As the raw impact of AI on artists’ lives became apparent many wanted to fight back. Within weeks artists were asking ‘what can be done to stop generative AI?‘. Within months, lawsuits had been filed against the major AI companies.
The Concept Art Association had raised over $270,000 through crowdfunding to hire a lobbyist to advocate for creators in Washington D.C., and activists in Europe had formed the European Guild for Artificial Intelligence Regulation (EGAIR), an organisation that lobbies EU lawmakers on behalf of creators.
It’s now been two years since the lawsuits against AI companies began to mount up, and since then political lobbying has got under way both in Europe and the US. So what progress has been made in this time?
Fighting back against gen AI
The first thing is that the lawsuits have stuck around, despite AI companies’ protestations that they should be thrown out. Many legal cases are ongoing, and the outcome of these will determine what material can be used to train AI models, and under what conditions.
Most of the cases allege that use of a plaintiff’s material without their consent constitutes copyright infringement. The AI companies argue that their actions fall within the “fair use” exception to the law.
It’s up to the courts to decide how existing copyright law will be applied to AI, but these cases are making progress, and the fact they’re seen by judges as having sufficient merit to proceed is promising.
One of the most important cases in the US has been filed by a group of artists against a string of companies for their use of Stable Diffusion. Follow its progress at Image generator litigation. In the UK, one to watch is Getty Images v Stability AI.
Another arm of anti-generative AI activism is the lobbying of governments and lawmakers, and in Europe, lots has been achieved in this area. The European Union (EU) has approved the AI Act, the world’s first comprehensive law relating to AI.
The Act requires AI companies to obtain consent from rights holders before using their material to train models; to label AI-generated content as such; and to provide a publicly available summary of the data training data.
Are AI companies breaking the law?
EGAIR was instrumental in achieving these victories, and their work is far from over as AI companies will seek to influence the final draft of the Act as it passes through the last stages of becoming law.
Looking back over the last two years, EGAIR’s founder Francesco Archidiacono reflects on how things have changed.
“Two years ago, talking about companies illegally using our data, infringing copyright, and the right to privacy was more complicated,” he says. “The politicians and institutions weren’t particularly informed on what was happening and much of the discourse was still centred on whether copyright had anything to do with AI training.”
But today, things are different. Francesco adds: “We’ve had OpenAI publicly admitting their need for high-quality and copyright-protected content for their AI services, the EU AI Act explicitly states that any use of copyrighted content to train an AI model must be authorised by the rights holder, and politicians are usually much more sympathetic to our requests.”
He adds: “I’ve been in more than one meeting in Brussels where I was told it was clear that certain AI companies have broken the law. It has become much easier to explain the issues with AI companies without looking like crazy anti-technology weirdos.”
Francesco also sees the reaction of some AI companies to the new EU laws as a sign that they’re on the back foot. The transparency requirements around labelling AI content and providing information about the data used to train models have been a sticking point for some companies.
“Despite the light weight of this burden, we’ve seen a lot of companies complaining to EU authorities, and Meta deciding not to release their generative AI services on the European market,” says Francesco. “This tells me that some of them are scared of what complying with the European Union’s transparency requirements might reveal.”
On top of lobbying and legal action, a key priority for activists is keeping the public and the art community informed – something that Francesco describes as “a big challenge”.
He says: “I fear nobody, us included, is doing a good enough job with that. We need to be better at teaching our peers what’s going on, and if we could find the resources and the time to do it properly, I genuinely believe that the fight would be much easier.”
The illustrator and creature artist Eva Toorenent is also EGAIR’s representative for the Netherlands. She believes that these days, public opinion is on the side of artists, and making sure people are informed will keep it that way.
“People are becoming more aware of AI’s enormous environmental impact, ethical concerns surrounding its use, and its other potential dangers, such as misinformation, fraud and deepfakes,” she says.
There is also a growing sense of distaste around AI-generated material. Eva tells me: “When I inform someone that an image is AI-generated, they often express feelings of being duped.
“Companies are updating their contracts to ensure their commissioned work is free from AI-generated material. I’ve also noticed a growing appreciation for human creators and craftsmanship.
“The more AI-generated images flood the market, the less I perceive it as a threat. In my opinion, the unending production of generative AI content has increased the value of human-made work.”
Eva contends that one of the most important things in this fight is that people know their rights.
She says: “What I’ve learned in my activism these past two years is that you, alone, can have more impact than you think.”
Many organisations have sprung up to help, and Eva urges artists to look for the ones in their own countries. In the Netherlands, an organisation called Opt Out Now! enables artists to opt their entire body of work out of AI training with a single application.
Eva also emphasises the importance of calling out companies and people who use unethical generative AI. She says: “For instance, I’ve noticed several artists using generative AI to create images that they then copy exactly, even replicating the mistakes.”
Eva addS: “In mimicking Midjourney images, they lose their unique artistic voice and style. Artists using this technology in this way take advantage of their peers and fellow creatives. We should resist normalising the use of a technology that’s fundamentally rotten. Solidarity among creatives is crucial.”
It’s easy to get disheartened by generative AI and the other threats to the creative community that exist today. We’ve heard from Stanley “Artgerm’ Lau on his fears human artists could be replaced. but Eva tells us that one of the most powerful things we can do is to keep creating.
“Art is human. Everything in the world has been created before, but not yet by you. Art is so much more than just a product. It’s a reflection of who you are as a human.
“It would be a shame if you would deprive the world of your unique vision because some thieves can’t keep their hands to themselves. Human beings and art are flawed, complex and strange. That’s what makes them interesting.
“I don’t think people want to see movies, art and music made by computers. Because who do you want to support? AI companies that steal our work, mutilate it, and sell it back to us? Or the person who has spent years developing a unique voice?”
Source: These digital artists have made progress in the fight against unethical AI Creative Bloq
