AI Law
AI Regulatory Environment: Latest Developments
Author
Barbora Korcova
Author
Barbora Korcova
February 13, 2024
AI Law
Software & Technology
The AI sector currently contributes £3.7 billion to the UK economy and employs 50,000 people across the country. The AI industry is thriving, and the impact of artificial intelligence on UK economy is set to increase dramatically.
To address the fast development of AI technologies and prevent risks associated with it, there is pressure on the UK to act quickly in the area of AI regulation to avoid being left behind other AI global leaders like the EU, US or China.
What is the Government doing in the field of AI regulation
Artificial intelligence generally, like any technology, is in part regulated under a range of pre-existing regulation such as data or consumer protection laws, but there is currently no AI specific legislation in the UK.
Therefore, in March 2023, the Government Department of Science, Innovation & Technology (DSIT) published its AI White Paper with the objective to drive growth and prosperity, increase public trust in artificial intelligence and strengthen the UK’s position among other AI global leaders. The Government decided to adopt a non-statutory, pro-innovation and principle-based approach.
It is important to point out that the White Paper is not proposing to introduce an AI law such as EU’s AI Act (the world’s first comprehensive AI legislation expected to become EU law in early 2024). Instead of creating a new AI regulator, the aim is to empower existing UK regulators to create tailored guidance to deal with issues relevant to their sector. The effectiveness of this non-statutory framework will be monitored and if considered necessary, the Government will introduce a statutory duty on regulators to have due regard to the principles below.
The proposed AI framework will be underpinned by five cross-sectoral principles for the UK’s regulators to apply within their remits. These principles follow OECD’s Principles for Ethical AI Use and are:
- safety, security and robustness;
- transparency and explainability;
- fairness;
- accountability and governance; and
- contestability and redress.
Government Response to the AI White Paper consultation
Following the publishing of the White Paper, the Government conducted a public consultation between March and June 2023. After the consultation concluded, the Government published its Response to the results of the consultation on its White Paper.
Some relevant points from the Response:
Support for principles and next steps for regulators
The responses to the consultation showed a strong support for the cross-sectoral principles. Some regulators are already proactively implementing these principles, for example the Information Commissioner’s Office (ICO) updated their guidance on how data protection laws apply to AI systems to ensure fair processing of personal data.
Now, the Government asked relevant regulators to outline their approach to AI White Paper by 30 April 2024. As indicated in the White Paper, the principles will be implemented on a non-statutory basis and that confirms the Government’s plan to introduce a voluntary regulatory regime for artificial intelligence.
No code of practice on copyright and AI
There will not be a code of practice on copyright and AI as the Intellectual Property Office (IPO) and a working group on artificial intelligence could not agree on it.
However, the Government wants to work with AI developers and copyright holders to ensure that the artificial intelligence and creative sector grow together. Further proposals of how this will be done will be published soon.
AI Regulation Roadmap
The Government set out the next steps for 2024, including working to develop AI regulatory policy to tackle AI risks, identifying existing gaps in existing regulatory powers, developing the central function within Government to coordinate regulatory activity and support international collaboration on AI Governance.
What is ICO doing in the field of AI
We are also keeping a close eye on the ICO consultation series on generative AI and data protection law.
ICO guidance, referred to above, is not expressly binding on organisations that process personal data (also true of the ‘Recitals’ to the UK GDPR) in the same way as the UK GDPR itself. However, any Court, and indeed the ICO itself, would pay close attention to the ICO’s interpretation of the UK GDPR in any enforcement of the relevant provisions at a later date.
The ICO recognises the necessity of addressing potential issues that may arise with the use of generative AI models such us ChatGPT in respect of data protection law. The ICO has released the first chapter of its consultation series, focusing on the lawful basis for training generative AI models using web-scraped data.
Interestingly, and very relevant to any generative AI developer, is that the ICO acknowledges that legitimate interests can serve as a lawful basis for using web-scraped personal data to train generative AI models, provided that the usual ‘legitimate interests’ test is satisfied (purpose, necessity and balancing test).
The focus in its preamble to the first chapter of the consultation is mainly on the balancing test, outlining several risk mitigations for developers of generative AI to consider. These include implementing ongoing technical and organisational measures, establishing monitoring processes and imposing contractual controls over third parties.
The ICO specifies three key aspects that developers using web-scraped data to train generative AI models must be able to demonstrate: a valid and clear interest, careful consideration of the balancing test and a demonstration of how the identified interest will be realised, along with mitigation of risks.
This is all existing law, however, to understand in more detail the ICO’s own view is extremely useful in seeking to comply with the UK GDPR.
The initial consultation on the first chapter will conclude on 1 March and the ICO plans to publish additional chapters in the coming months, which we will be closely monitoring.
Comment
The use of artificial intelligence is growing across all sectors, and this presents challenges for UK regulators.
The UK Government’s proposed AI framework is designed to be flexible and based on the collaboration between government, regulators and businesses. Arguably, there might be a need for greater clarity (in a form of legislation) in case of discrepancies between regulators and their sectors. It is possible that with the evolution of the artificial intelligence, the Government will decide to change its current approach and legislate, which, considering that artificial intelligence is a very fast-moving field, will be challenging as a set statutory framework may struggle to adapt to its developments.
The initial failure of the IPO to agree on a copyright code of practice goes some way to demonstrate just how complicated this issue is. We’ve discussed elsewhere how potential web scrapers face significant legal complexities around copyright in another article, and artificial intelligence (which can rely heavily on web-scraping techniques) suffers from similar problems.
Note: we bring to your attention the only IP infringement case in the UK that relates to generative AI: a lawsuit filed by Getty Images against generative AI provider Stability AI, alleging that Stability AI unlawfully used its images to train its AI model and therefore committed a copyright infringement. The High Court ruled that this case can proceed to trial, and it will be interesting to follow how the Court deals with this case that will set a precedent for future cases involving the use of generative AI.
The complexity and unpredictability of artificial intelligence makes its regulation troublesome. While the Government chose to adopt a non-statutory approach to regulation in the UK for now, the ICO is publishing guidance to help businesses stay compliant with already existing data protection laws in the age of artificial intelligence.
Although the Government avoids placing regulatory burden on businesses for now, forward-looking businesses might wish to consider implementing policies governing use of artificial intelligence and its development and align them with the five principles as well as keeping an eye on the ICO that will keep publishing guidance on artificial intelligence in relation to data protection laws.
What’s next?
At EM Law, we specialise in technology law. If you have any questions about the Government White Paper on AI, the ICO guidance or AI more generally, please get in touch with Colin.
