shutterstock_619078730_melinda_nagy
6 February 2023CopyrightMuireann Bolger

Reaction: UK scraps radical AI mining proposals

Lawyers say the country is now behind the EU and Japan | Government may have pushed key AI questions to the courts | Finnegan | Marks & Clerk | Bird & Bird | Brown, Neri, Smith & Khan | Gowling WLG.

The UK government has abandoned controversial plans to allow AI developers greater powers to mine text and data, following an outcry from copyright owners.

Last year, the June 2022 Government response to a consultation on Artificial Intelligence and Intellectual Property proposed a new copyright and database exception that allows text and data mining for any purpose.

That month, the Publishers Association issued a letter arguing that the government should rebalance its proposal in favour of copyright owners.

The proposal was later rejected by the House of Lords which branded it “misguided” and warned that these changes “take insufficient account of the potential harm to the creative industries”.

In response to concerns, UK digital minister Julia Lopez gave evidence to a parliamentary committee on January 25, stating that: “ IP is the lifeblood of many creative industry businesses, so we are not convinced of the value of this piece of work. I am fairly confident, in so far as I can say publicly, that this is not going to proceed.”

The development comes as Getty Images has sued generative AI developer Stability AI for infringement, and as generative AI developers are facing a swathe of infringement lawsuits.

In a tweet, Liberal Democrat MP Sarah Olney argued that a “text and data mining exception to UK copyright law would have left those in creative industries without legal control over their work, should it have been used by an AI platform”.

She welcomed the development, describing it as “a huge win for millions employed in creative spaces across the UK...”

Following the development, WIPR asked IP and AI experts for their views and what might happen next…

Matt Hervey, head of AI, Gowling WLG: “We’re back to square one—and behind the EU.”

“Undoubtedly, outputs of the "generative AIs" launched in 2022 are remarkable—they are good enough to win awards, to be used for magazine covers and to be accepted into stock image libraries.

“The power of current generative AI depends on vast training sets—in some cases the whole ‘crawlable’ internet, eg, GPT-3 was trained using multiple databases of text including Wikipedia, books hosted on the internet and the ‘CommonCrawl’, a database of text extracted monthly from billions of web pages.

“In June, the Government announced a plan to allow mining ‘for any purpose’. That would probably have been the widest exception in the world, but by November it was clear the government had changed its mind.

“Like all jurisdictions, the UK needs to strike a balance between the interests of copyright owners and technological innovation. The answer will be a mix of practical, moral, social, economic considerations and politics. This is a key area to watch, debate and influence.

“There's no ‘right’ answer on how broad the UK's text and data mining exception should be. But it does not help that the debate has become so polarised and that no serious economic impact assessment has been published since 2015.

“Moreover, the government's proposed exception ‘for any purpose’ needed to be backed up by details of the potential safeguards, such as the limits of ‘legal access’ and whether AI outputs could be infringements. Without those, it's not surprising we're back to square one—  and behind the EU.”

Ryan Abbott, partner at Brown, Neri, Smith & Khan: “A hot policy issue in the UK.”

“Whether and to what extent a statutory exception is adopted for text and data mining for copyright infringement is a critically important issue for both AI developers and rights owners.

“The exception question will continue to be a hot policy issue in the UK and internationally for years to come, and will only become more important to resolve as AI capabilities and adoption continue to improve.”

Toby Bond, partner, Bird & Bird: “EU and Japan more permissive.”

“Once a leader in the AI space, the UK’s 2014 non-commercial research exception has been overtaken by more permissive approaches in jurisdictions such as the EU, Japan and Singapore.

“The decision last summer to move ahead with a broad text and data mining exception was presented as making the UK a location of choice for data mining and AI development and supporting the government’s ambition for the UK to be a global leader in AI innovation.

“However, rights owners have clearly pushed back against this, likely citing rapid advances in generative AI since the decision was made. The key for all involved will be what happens next, and how quickly. While we wait to find out, AI developers will continue to make decisions about whether to invest in the UK.”

Darren Jiron, managing partner, Finnegan: “Not extending the exception may be overly limiting.”

“Not expanding the exception seems aimed at the possibility that the trained AI systems would generate a reproduction of at least a part of a copyrighted work.

“While this is possible (eg, requesting that a trained AI system generate a composition including a segment extracted from a particular song or work of art), much of the focus of trained AI systems is on a new capability to generate an original work conceived of by a human user of the trained system.

“A request to generate a drawing of the Tower of London in the style of Monet may be based on the network’s previous exposure to paintings by Monet, but in most cases, the generated work will not include segments reproduced from any Monet painting.  In view of such cases, not extending the exception may be overly limiting.

“Training an AI system through exposure to previous works of art is much like training a human student through similar exposure. Asking either the trained AI system or the trained human student to generate an original work of art based on experience likely results in an original work protected by the copyright laws rather than an infringement of rights of other artists.”

Mike Shaw, partner, Marks & Clerk: “UK government may have simply pushed the question… to the courts.”

“Since the creation of a temporary copy of a copyright work constitutes copyright infringement unless the temporary copy is for enabling ‘a lawful use’ of the work, it is interesting to note that the government has declined to address whether the creation of temporary copies for the purpose of training AI should constitute lawful use.

“This demonstrates the challenge in finding a balance between the interests of AI developers and the creative industries, particularly in light of the value that these industries play in the economy.

“The question is: to what extent should copyright owners benefit from use of their works by AI, and how should AI developers be able to benefit from the original creations of copyright owners?

“There are options that could go some way to addressing these challenges, such as an opt-in scheme whereby copyright owners receive royalties for use of their works in AI training, but we’ve seen how difficult it can be to implement such compromises before, eg, the attempts by the EU Copyright Directive to impose obligations on content-sharing platforms to monitor copyright materials uploaded to the platforms without the copyright owner’s consent.

“By evading the issue the government may well have simply pushed the question of whether AI training amounts to ‘lawful use’ of a copyright work onto the courts to decide, and it remains to be seen whether the judiciary will be willing to answer this question.”

Did you enjoy reading this story?  Sign up to our free daily newsletters and get stories sent like this straight to your inbox.

Today’s top stories

‘No more First Amendment smokescreen’ for NFTs?

MetaBirkins creator loses to Hermès in landmark NFT case

Already registered?

Login to your account

To request a FREE 2-week trial subscription, please signup.
NOTE - this can take up to 48hrs to be approved.

Two Weeks Free Trial

For multi-user price options, or to check if your company has an existing subscription that we can add you to for FREE, please email Adrian Tapping at atapping@newtonmedia.co.uk


More on this story

Copyright
20 July 2023   In a recent webinar, the US Copyright Office clarified when applicants need to disclose AI contributions to their work, say Mark Doerr and Alexis Marin of Greenspoon Marder.
Patents
24 August 2023   Series marks, mediation, and online documents included in proposals | Consultation marks second phase of One IPO Transformation Programme to enhance IP office’s digital services.
Copyright
6 September 2023   Government publishes five objectives for the Bletchley Park event | No specific mention of IP or protection for creators.