The world’s leading photo agencies and photographer associations have co-signed an open letter calling for legal protections against artificial intelligence (AI).
Getty Images, the Associated Press, Agence France-Press, the European Pressphoto Agency, and the National Press Photographers Association are among the organizations calling for intellectual property (IP) rights to be respected.
While praising generative AI technology and its potential benefits to society, the signees warn that a flood of synthetic content into the public sphere has the potential to undermine the public’s trust in the media.
Tellingly, the letter refers to “violating copyright law” and demands that AI companies get the consent of IP rights holders for the “use and copying of their content in training data and outputs.”
The companies also advocate for transparency — “as to the makeup of all training sets used to create AI models” and for media companies to “collectively negotiate with AI model operators and developers regarding the terms of the operators’ access to and use of their intellectual property.”
The picture agencies also want AI models and users of the technology to “clearly, specifically, and consistently identify their outputs” as AI-generated content. As well as “requiring generative AI model providers to take steps to eliminate bias in and misinformation from their services.”
The letter calling for AI regulation can be read in full here.
AI models are built on giant scrapes of the internet. Meaning, generative AI companies effectively mined what they call “publicly available” data. E.g. images and text available on Google, social media, and elsewhere on the internet.
However, as most people are aware, you cannot just take someone’s photo from Google Images and start using it for any reason you want, that photo generally belongs to someone.
If your photo has ever been online, there is a strong chance that it has been used by an AI company to train their model with. There is even a website set up to search one of the major data sets that AI companies use.
Getty Images is already suing Stable Diffusion for using 12 million of its copyrighted photos that were in the LAION data set. AI image generators like Midjourney and DALL-E haven’t revealed their data sets.