News Over 250 UK publishers opt in to collective AI licensing scheme

[Claire G]

The announcement, which comes ahead of the annual PLS Conference on 2nd July, shows industry support for an opt-in collective licensing scheme that gives publishers the opportunity to safeguard and monetise their work.

According to the PLS, the industry is now a “significant step closer” to offering a licence and online content store for AI companies to legally access and use published works in exchange for a licence fee, under clear and transparent usage guidelines.

The collective licence scheme, which has been developed with the Copyright Licensing Agency (CLA) and the Authors’ Licensing and Collecting Society (ALCS), has been designed to support publishers who may be unable to secure direct licensing agreements with AI developers by providing them with an optional alternative route to market that allows them to license their content for generative AI.

Any thoughts? I know not every has a subscription to The Bookseller, but for those who do here's the link to the full article:
Over 250 UK publishers opt in to collective AI licensing scheme

@AgentPete would authors have a say in this? And do you have any thoughts on what 'clear and transparent usage guidelines' might include?

 

[RK Wallis]

The announcement, which comes ahead of the annual PLS Conference on 2nd July, shows industry support for an opt-in collective licensing scheme that gives publishers the opportunity to safeguard and monetise their work.

According to the PLS, the industry is now a “significant step closer” to offering a licence and online content store for AI companies to legally access and use published works in exchange for a licence fee, under clear and transparent usage guidelines.

The collective licence scheme, which has been developed with the Copyright Licensing Agency (CLA) and the Authors’ Licensing and Collecting Society (ALCS), has been designed to support publishers who may be unable to secure direct licensing agreements with AI developers by providing them with an optional alternative route to market that allows them to license their content for generative AI.

Any thoughts? I know not every has a subscription to The Bookseller, but for those who do here's the link to the full article:
Over 250 UK publishers opt in to collective AI licensing scheme

@AgentPete would authors have a say in this? And do you have any thoughts on what 'clear and transparent usage guidelines' might include?

Oh, Clare, you're depressing me, lol

 

[AgentPete]

@AgentPete would authors have a say in this? And do you have any thoughts on what 'clear and transparent usage guidelines' might include?

This is actually what needs to happen, i.e. the effective (and volitional) licensing of writers’ works for use in, or exclusion from, LLMs (Large language Models).

The situation is very fluid and undefined at the moment. EG, generally under UK law, copying a book to use to train a LLM is a restricted act under the Copyright, Designs and Patents Act 1988... with key exceptions.

Aggressive lobbying by the limitlessly-funded AI industry resulted in the British government initially saying that because the AI industry was so “important”, it could have a broad exception to the Copyright, Designs and Patents Act... but authors could also have the right to opt out from having their work freely cannibalised and pirated if they wanted to. Big deal!

However, things now seem to be moving against the government’s initial position, and today’s announcement is one indication of this. But the UK battle is very far from won.

In the US, the situation is less promising. “fair use” and “transformative use” have given the big boys carte blanche to plunder our work without any compensation. One slight ray of hope: while scanning a book for training a LLM is considered to be :transformative: and therefore fair use, creating a digital library of pirated books is not fair use. Anthropic (Claude AI) did exactly that, and now have to pay authors statutory damages.

In practice, because the US law gives maximum protection to the AI companies, this means that books travel to the US to be scanned and input into their LLMs.

So... how does this announcement affect authors?

It doesn’t yet. It simply means there is a sort of framework developing for UK licensing.

Whether the AI companies will play ball is entirely another matter.

They’re all losing money at the moment. Adding yet another cost to their operations is not going to be welcome.

But yes, eventually this may filter down to an item in an individual writer’s publishing contract. A that point, as I’ve said before, the key issues will be:

• How much are your being paid for those rights?
• Over what time scale and territory?
• How can you withdraw permission and/or renegotiate the agreement after x years?
• What performance bonus is there (e.g. payment by usage)
• And most importantly, because AI companies are essentially pirates, how can we check that they are honouring the contract anyway?

p.

 

[Claire G]

This is actually what needs to happen, i.e. the effective (and volitional) licensing of writers’ works for use in, or exclusion from, LLMs (Large language Models).

The situation is very fluid and undefined at the moment. EG, generally under UK law, copying a book to use to train a LLM is a restricted act under the Copyright, Designs and Patents Act 1988... with key exceptions.

Aggressive lobbying by the limitlessly-funded AI industry resulted in the British government initially saying that because the AI industry was so “important”, it could have a broad exception to the Copyright, Designs and Patents Act... but authors could also have the right to opt out from having their work freely cannibalised and pirated if they wanted to. Big deal!

However, things now seem to be moving against the government’s initial position, and today’s announcement is one indication of this. But the UK battle is very far from won.

In the US, the situation is less promising. “fair use” and “transformative use” have given the big boys carte blanche to plunder our work without any compensation. One slight ray of hope: while scanning a book for training a LLM is considered to be :transformative: and therefore fair use, creating a digital library of pirated books is not fair use. Anthropic (Claude AI) did exactly that, and now have to pay authors statutory damages.

In practice, because the US law gives maximum protection to the AI companies, this means that books travel to the US to be scanned and input into their LLMs.

So... how does this announcement affect authors?

It doesn’t yet. It simply means there is a sort of framework developing for UK licensing.

Whether the AI companies will play ball is entirely another matter.

They’re all losing money at the moment. Adding yet another cost to their operations is not going to be welcome.

But yes, eventually this may filter down to an item in an individual writer’s publishing contract. A that point, as I’ve said before, the key issues will be:

• How much are your being paid for those rights?
• Over what time scale and territory?
• How can you withdraw permission and/or renegotiate the agreement after x years?
• What performance bonus is there (e.g. payment by usage)
• And most importantly, because AI companies are essentially pirates, how can we check that they are honouring the contract anyway?

p.

Thanks @AgentPete Very comprehensive and lots of food for thought xx