We may no longer be part of the European Union but the UK may be influenced or affected by other nation's laws, especially on global issues such as climate change and online safety. One such area currently provoking concerns is the advancement of Artificial Intelligence technologies.
Many businesses and even members of the public are adopting AI usage into their daily lives or work operations. Tools such as ChatGPT have spurred conversation and debate not only as to the potential of AI technology but also the threat. I wrote an article in November 2023 addressing the potentials, the pitfalls and the potential pitfalls regarding the use of ChatGPT in the workplace and what was most concerning is we have no legal framework for how we are going to control, manage and potentially limit the use of these technologies. This means misuse of AI, or cases in which AI usage results in detrimental mistakes, holds no one specifical accountable.
Already there have been numerous legal cases and failures of AI that have led to data leaks, IP infringement and false information being used. Currently, these are being resolved on a case-by-case basis but ultimately we’re going to need to do what we did with data protection and create a legal framework. Just recently though, the European Parliament got ahead of us and introduced a landmark Artificial Intelligence Law that seeks to balance innovation with fundamental rights and protections.
Why is the EU’s New AI Law Important?
Since the internet is a global entity, any legislation created anywhere in the world has the potential to impact other countries. For businesses, this is particularly important because compliance with the laws in the countries you operate in is required. This means if citizens of other nations can access your website, services, or products in their country then it/they must meet with the legislation in the destination location, as well as the country of origin.
Moreover, the UK will likely implement its own legislation around AI - hopefully after the UK Data Protection Laws have been updated - and this is expected to take a similar approach to the EU.
What Is In The EU Artificial Intelligence Act?
The Artificial Intelligence Act was passed by the European Parliament and came into force on August 1st, 2024. Its aim is to protect fundamental rights and existing laws without compromising innovation and the ability of AI to positively impact our lives.
The EU Artificial Intelligence Act is a complex document but there are three key areas of the legislation, comprising of the following:
Reducing Risk
The EU AI Act seeks to ban technologies that -
Seek to manipulate the behaviour of people, particularly vulnerable groups such as children
Classify people based on behaviour, social economic circumstance or personal characteristics
Use biometric identification or categorisation systems including facial recognition, emotion recognition (in workplaces and schools) social scoring, predictive policing, etc.
These restrictions will exist in the private and public sectors but may - only in very specific circumstances - be lifted for matters of national security and law enforcement.
Restrictive Use
AI may not be applied as a reliable source of information in certain sectors and for certain applications. These include -
Critical infrastructure
Education (including vocational)
Employment and access to employment or self-employment
Law enforcement
Access to essential services (both public and private)
Border control, migration and asylum applications
Assisting in legal matters including interpretation and application of law
Disclosure
Use of AI technologies must be published and must also adhere to existing laws, especially around data protection and intellectual property. General-purpose AI systems need to meet transparency requirements and be compliant with copyright law. Summaries of source training materials must be detailed and published.
In addition, artificial or manipulated images, audio and videos must be clearly labelled as such, in the same way that advertisements on social media are required to tag themselves.
The Future Of AI Legislation
It’s highly important that the law catches up with the huge leaps in technological advancements that have occurred in the past decade and, being that AI could have a profound impact on our lives and the future of work, this piece of legislation is a step in the right direction. Unfortunately, it’s not applicable to the UK but we may see similar legislation make its way into UK law within the next year or two.
Measures seem to have been taken within this legislation to tackle some key concerns, some of which have even been flagged by AI creators, including the potential for systems to apply bias in decision-making.
Copyright infringement is also addressed, although other countries, such as the USA, have quite different legal processes and standards around copyright which may affect how an AI might be trained depending on the country it is created in. Potentially, this may restrict where materials and data used to train AIs are sourced from. So, legislation around AI is not an uncomplicated problem. I am glad to see it is beginning to be properly addressed though. I endeavour to keep you updated!
This blog was written based on the legal position as at 26th August 2024.
Comments