Insights

AI text and data mining

Government U-turn aims to protect creative industries from AI text and data mining

The government has changed its mind over whether to allow those developing AI systems to have the right to use any content and data they want in training their AI.

We take a look at the background and the concerns of those in the creative industries over the use of their content to create and teach AI systems.

Training AI systems

AI systems mine a vast amount of text and data to use in training. While some datasets and databases are freely available, other online content may be protected under copyright and database protection rules.

This means that when an AI training system copies data and extracts information that is legally protected to create new works, it will be infringing legislation where there is no licence or legal exception.

Current exceptions include using data for non-commercial research and private study.

Proposed copyright and database right exception

In July 2022, the government carried out a consultation on AI and intellectual property rights. It proposed an exception to data protection and copyright legislation which would allow text and data mining (TDM) to be carried out on content otherwise protected under intellectual property rights laws. Those with intellectual property rights over the work would not have any option to object to the exception or to opt out of it.

Creative industry concern

Creative industries, including the music industry, raised serious concerns that AI would be able to use content with no consent from those who had worked to create it and no compensation to them.

UK Music’s chief executive, Jamie Njoku-Goodwin wrote a letter to the Culture Secretary in which he said that the proposal “would significantly undermine the basic principles that the creative industries are based on.”

He said: “Our innovation has value. Our creative IP is not raw material for others to freely monetise – yet this is what these proposals open the door to.

UK Music supports attempts to grow the UK’s AI sector, yet this cannot be achieved by taking away vital tools that enable the music industry, and other IP-reliant sectors, to innovate.

These proposals are dangerous to the future prospects of our globally successful sector. As someone who has always championed our world-leading creative industries and knows from personal experience just how important robust copyright protections are, we ask for your support to help to guarantee this proceeds no further.”

Government change of heart

In light of widespread concern, the House of Lords Communications and Digital Committee cast doubt on whether the government’s proposed consent to unlimited TDM by those training AI should be given.

Then in February 2023, the former Minister for Science, Research, and Innovation, George Freeman, confirmed that the proposal to make it legal to allow data and other content to be scraped regardless of copyright would not be proceeded with.

This means that for now, rights holders can rely on existing rules and those training AI systems will need to take care not to use protected content and databases.

Future changes

Legislation is still likely to be enacted at some point to deal with intellectual property rights and how these will work alongside the creation of AI. The challenge will be to balance the interests of rights holders with those creating AI and create a system that is both fair and enforceable.

Contact our company and commercial solicitors in Westminster, London

Martin Donoghue is a partner in the Company and Commercial department. One of his main focus areas is technology.

If you would like to speak to Martin or one of his team, call 020 7873 1000, email law@williamsturges.co.uk or fill in our contact form below.

Share