The AI Opportunities Action Plan, commissioned by the UK government and written by Matt Clifford, was published today. It’s a 25-page document that outlines a plan for how the UK can “embrace the transformative potential of AI” in a bid for economic growth, in the form of 50 recommendations.
Brilliant commentary, Ed. I’ll also be quoting you and Graham Lovelace in an article I’m writing. Let’s also not forget the rhetoric vs. reality macroeconomic backdrop of this Action Plan - grandiose plans and no specifics about how all this will be funded. It’s a tech-giant heist that’s full of A-irony. Let’s keep banging the drum as I know a lot of creatives who seem blissfully unaware that their work is about to be raided.
"The market for licensing training data is thriving, and a number of AI companies already license all of the work they train on."
Could you provide some examples of such companies?
Provided that they are not companies that also operate stock image services - be that GETTY, ShutterStock, Adobe (Firefly) - where said licensing is considered dubious by many, are their models competitive with models from companies that disregard such licensing requirements?
nice .. we are not going to continue drawing so that the founders of ai can earn, get rich and work for us .. then create it yourself .. I am quitting this job now
It should be noted that all of society is affected by this, not just the creative industry. Copyright protection is granted by default, no declaration needed, to every single work of intellectual and original origin, including any written text, such as social media posts, pictures, business documents, web pages, advertisements, product descriptions, books, software, technical documents, family albums, company brochures, logos, any kind of graphic, presentations etc. Removing default copyright protection for the benefit of a few will upend all the very fabric of the modern world.
The consultation isn't even supposed to end until next month, and this is how they're already acting?
This goes beyond just discourse surrounding ethics in artificial intelligence, this is a disregard for the democratic process itself. They're already caving into corporate interests and ignoring the demands of the public, including workers in industries that would be negatively effected by their proposed changes. In a time where people all over the world are losing faith in democratic institutions, moves like this send the message that the government does not care about the people at all. Not unless they have the backing of big industries, at least.
I’m a supporter of strict copyrights, but it has become clear to me that some compromises might be surprisingly beneficial.
Think about tax evasion. Unless we can get all governments in the world to agree on something, offshoring can only make things worse. Imagine a world where data are mined in Bermuda, models trained in Panama, parameters washed in Singapore, and products sold to Europe.
A much better yet sub-optimal alternative would be new regulations. I am not sure whether an opt-in or opt-out would be the best and in which form they should be done. But, we certainly need to do something before companies turn to “black markets”.
And think about digital content. It is a paradox or even a contradiction in my opinion. Something “free” to share and distribute while asking a price to sell. I understand many creators are relying on viewing/streaming numbers to make a living. But, isn’t it a road to nowhere itself? It is a model based on popularity but the value of said work. And people are complaining about the rise of populism. The society needs change, the internet needs change. Otherwise, copyrights have to change.
Finally, I agree the author is doing an amazing job to tame the companies to do good. Regulations cannot solve every issue. We need to be conscious of what we consume and it is a good way to get started.
Disclaimer: I’m an artist with tech background and my work can’t be reproduced/copied by current AI in any means
Excellent post. However, where in the EU AI Act does it provide an opt-out option for AI training? Afaik the EU AI Act requires transparency on the data sources used for general purpose models but does not grant special rights for training.
It does, and it requires them to be in "machine readable format". Stock photographer Kneschke takes non-profit dataset provider Laion through the hoops in the Hamburg court to find out if file metadata, site terms of service and on-page rights reservations are enough, or if new standards are required. Regardless, it's not practical. The default in copyright law is that all rights are reserved, and any and all content will be pirated, torrented, uploaded and orphaned...
Brilliant commentary, Ed. I’ll also be quoting you and Graham Lovelace in an article I’m writing. Let’s also not forget the rhetoric vs. reality macroeconomic backdrop of this Action Plan - grandiose plans and no specifics about how all this will be funded. It’s a tech-giant heist that’s full of A-irony. Let’s keep banging the drum as I know a lot of creatives who seem blissfully unaware that their work is about to be raided.
Excellent post Ed ... I'll be quoting you in my coverage soon.
"The market for licensing training data is thriving, and a number of AI companies already license all of the work they train on."
Could you provide some examples of such companies?
Provided that they are not companies that also operate stock image services - be that GETTY, ShutterStock, Adobe (Firefly) - where said licensing is considered dubious by many, are their models competitive with models from companies that disregard such licensing requirements?
Absolutely - all the companies listed here: https://www.fairlytrained.org/certified-models
nice .. we are not going to continue drawing so that the founders of ai can earn, get rich and work for us .. then create it yourself .. I am quitting this job now
It should be noted that all of society is affected by this, not just the creative industry. Copyright protection is granted by default, no declaration needed, to every single work of intellectual and original origin, including any written text, such as social media posts, pictures, business documents, web pages, advertisements, product descriptions, books, software, technical documents, family albums, company brochures, logos, any kind of graphic, presentations etc. Removing default copyright protection for the benefit of a few will upend all the very fabric of the modern world.
The consultation isn't even supposed to end until next month, and this is how they're already acting?
This goes beyond just discourse surrounding ethics in artificial intelligence, this is a disregard for the democratic process itself. They're already caving into corporate interests and ignoring the demands of the public, including workers in industries that would be negatively effected by their proposed changes. In a time where people all over the world are losing faith in democratic institutions, moves like this send the message that the government does not care about the people at all. Not unless they have the backing of big industries, at least.
I’m a supporter of strict copyrights, but it has become clear to me that some compromises might be surprisingly beneficial.
Think about tax evasion. Unless we can get all governments in the world to agree on something, offshoring can only make things worse. Imagine a world where data are mined in Bermuda, models trained in Panama, parameters washed in Singapore, and products sold to Europe.
A much better yet sub-optimal alternative would be new regulations. I am not sure whether an opt-in or opt-out would be the best and in which form they should be done. But, we certainly need to do something before companies turn to “black markets”.
And think about digital content. It is a paradox or even a contradiction in my opinion. Something “free” to share and distribute while asking a price to sell. I understand many creators are relying on viewing/streaming numbers to make a living. But, isn’t it a road to nowhere itself? It is a model based on popularity but the value of said work. And people are complaining about the rise of populism. The society needs change, the internet needs change. Otherwise, copyrights have to change.
Finally, I agree the author is doing an amazing job to tame the companies to do good. Regulations cannot solve every issue. We need to be conscious of what we consume and it is a good way to get started.
Disclaimer: I’m an artist with tech background and my work can’t be reproduced/copied by current AI in any means
Fight back, using free software like, "Glaze," or, "Nightshade." Both digitally poisons artwork, or graphics in a way that corrupts Generative AI.
Thanks for this enlightening article
(depressing, but enlightening)
Excellent post. However, where in the EU AI Act does it provide an opt-out option for AI training? Afaik the EU AI Act requires transparency on the data sources used for general purpose models but does not grant special rights for training.
It does, and it requires them to be in "machine readable format". Stock photographer Kneschke takes non-profit dataset provider Laion through the hoops in the Hamburg court to find out if file metadata, site terms of service and on-page rights reservations are enough, or if new standards are required. Regardless, it's not practical. The default in copyright law is that all rights are reserved, and any and all content will be pirated, torrented, uploaded and orphaned...