Is your website information ready for the new AI data framework?

Source: Howden

Author: Deke Adams – Business Development Manager (Howden)

Read time: 5 minutes

The respective fortunes and processes of the AI and creative industries grated against each other for years before the advent of generative AI – GenAI – further exacerbated relations between the two sectors.?

Freely-available GenAI tools from Open AI, Google, Microsoft, and others, have enabled pretty much anyone to generate ‘creative’ content – content that’s based on pre-existing inputs drawn from online sources.

The technology uses ‘models’ – processing and analytical software – that learn the underlying structures and patterns of the data that’s used to ‘train’ them, in order to produce new composite content.

For the creative economy, the concerns are manifold. Some AI has the potential to disrupt revenue streams for a range of branded products and services; from publishing and music to movies and gaming. Proprietary data swiped from company websites can also be used to train AI models designed for competitive commercial applications.

No surprises therefore that data owners on the web are left beyond peeved when their prized data assets – audio, video, text, images – are used without permission to generate new content value which they do not share and do not profit from.

Conscionable AI developers, meanwhile, want to avoid a court summons in what’s seen as a legal grey area. The datasets used to train AI models can draw in copyrighted content and rights-protected IP, leading to unresolved proprietorial disputes that might call into question established notions of lawful ownership and attribution.

The government reckons that the uncertainty around how copyright law applies to AI is holding back the UK’s AI and creative sectors from reaching their full growth potential. It makes it difficult for creators to control or be paid for the use of their work, while posing legal risks for AI developers that stifle investment, adoption, and competitiveness.

The new Copyright and Artificial Intelligence open consultation from the Intellectual Property Office sets out to come up with a ‘framework’ that addresses the application of UK copyright law to the training of AI.

Running until 25 February 2025, the consultation solicits views on a way forward that achieves the UK’s objectives for both its creative industries and the AI sector. Briefly put, these are:

• Supporting rights holders’ control of their content and ability to be remunerated for its use
• Supporting the development of world-leading AI models in the UK by ensuring wide and lawful access to high-quality data
• Promoting greater trust and transparency between the AI and creative sectors

The government aims to deliver these through ‘a package of interventions’ that include a way for rights holders to reserve their rights, enabling them to license and be paid for the use of their work as used for AI training.?

Alongside this, and more contentiously perhaps, the government proposes an ‘exception to support use at scale of a wide range of material by AI developers where rights have not been reserved’.

This approach would balance a right holder’s ability to seek remuneration while providing a clear legal basis for AI training with copyright material, HMG says, so that developers can train leading models in the UK while respecting the rights of rights holders.

Various voices from within the creative industry have reacted with a reasoned if somewhat wary response to the consultation and its ambits. The Creative Rights in AI Coalition – a recently-formed grouping of rights holders that includes publishers, authors, artists, music businesses, unions, and photographers – does not support the proposed new exception to copyright. It says that ‘the priority should be to ensure that current copyright laws are respected and enforceable’.

Time to secure all business data

While predicated on the interests of the creative industries – within which, by the way, HMG includes IT, software, and computer services firms as a subsector – the Copyright and Artificial Intelligence consultation and its eventual outcomes will have implications for all kinds of businesses across verticals.

Businesses should already be thinking about how it could impact the way they protect their published IP and other content assets – and how prepared they’re to respond to both approved and non-approved third-party AI-enabled appropriation of those assets. Companies don’t necessarily see what they publish into the public domain as constituting ‘data assets’ – but in a changing, AI-prowled digital landscape, they really should.

Take a look around most company websites, and you’ll find plenty of useful info not marked as copyright-protected and with no ownership rights asserted. You’ll also likely find downloadable PDFs and images that lack proper rights declarations.

What might seem like routine corporate factsheets could prove useful to rival businesses when folded into a dataset for training commercially-focused AI models. Price information, particularly, can yield value to a competitor to inform their own pricing plans, as well as ?sales and marketing strategies.

Just having a general copyright notice tucked in a corner of your homepage may not prove sufficient protection in the event of an illicit website scrape. Implementing some simple standard noticing practices could help minimise exposure to copyright infringement, IP misappropriation, and plain old plagiarism.?

They include ensuring that, where appropriate, a business’s proprietary data in the public domain prominently displays copyright notices and ownership rights assertions throughout, along with cautions against unsanctioned use.