Douglas McLachlan
- Partner
Whether you choose to embrace it or not, we have truly entered the age of artificial intelligence (AI).
While the long-term implications for society are unknown, what is clear is that organisations need to act now to consider the legal and ethical implications of using AI in their business.
Don’t be alarmed by the technical terms such as ‘machine learning’ and ‘generative AI.’
Firstly, you need to identify the type of AI tool you plan to use, understand how it works and get an idea of its limitations. The most important point is that AI is not infallible, and its output may not be 100% accurate. AI ‘hallucinations’ can be amusing, but you and your employees need to know why they happen, be willing to check sources and generally take everything the tools offer to you with a pinch of salt.
Also, because of the way they were trained, AI systems can be guilty of bias. Much of that bias comes from us – from the data we put on the internet. If the internet only shows images of middle-aged white men as CEOs, or women as nurses, then the AI may assume that to be a rule. Attempts to combat this have led to absurdities, such as AI image generators asked to create images of German WWII soldiers in uniform producing surprisingly ethnically diverse pictures.
There is a difference between using ChatGPT to answer general questions and using AI as a tool to sift through hundreds of CVs and decide who you should hire (if so, beware of bias – see above).
Different use cases pose different ethical and legal considerations.
Bias, discrimination, automated decision making or a lack of transparency in the AI system can all pose challenges when dealing with employees or potential recruits.
For example, lawyers who plan to use AI must consider client confidentiality and whether their clients are prepared to accept the output of the AI system without professional oversight (hint – not a lot).
Public sector bodies also need to consider a host of public law issues.
These matters require careful thought because using AI usually involves sharing data.
What data do you propose to upload? Where will the AI system process the data you put into it? Will the AI ‘learn’ from your data – if so, you can’t put anything confidential or private in there.
As a starting point, if you plan to put any intellectual property, or confidential, commercially sensitive or personal data into the AI system, then generally speaking your use of the system and your data needs to be walled off from the wider world and ideally held in the UK or EEA.
However, the wider point is that you are already on the back foot if your UK GDPR compliance is not up-to-date.
Few contracts will say “you may not use AI”, but many will say “you can only use the personal data we provide to you in this way” (and not mention AI). Many will require that personal data be kept in the UK or EEA.
In short, if you input personal data into a cloud computing AI system that runs from a data centre in Texas, you may be in breach of contract. You may also be breaking data protection law.
Guidelines for employees on the use of AI is a good start.
While this is a wider issue, it comes into sharp focus alongside the use of AI. You should develop a robust management plan for dealing with cyber and personal data breaches. This plan needs to sit outside the company servers in case these are compromised. Game plan a cyber attack that completely shuts you out of your computer systems, and work out what you need hosted elsewhere to get your business back up and running.
The contract may set out who owns the intellectual property rights in whichever content the AI tool creates for you based on your prompts, or it may contain important disclaimers and limitations of liability.
Your employees may already be using AI at work, as ChatGPT has a free – albeit limited – version.
You will need to move quickly to provide basic guidelines on the safe and proper use of AI in the meantime. The most important point here is to clarify what information they can and cannot upload into the AI system, and to set out what they can and cannot use AI for (see above).
Whether you’re just getting started on your AI journey or are about to procure an expensive AI system to revolutionise your operations, take legal advice from the outset to ensure your business is protected.