With the rise of platforms such as ChatGPT and increased investment in technology (particularly “deep learning” and “big data”), atificial intelligence (AI) has reached the stage of being able to produce content that, in some cases, is indistinguishable from that
generated by humans.

In June 2023, Forbes reported that AI contributed £3.7 billion to the UK economy in 2022. And who among us hasn’t used it to draft an email, a document, or even a contract?

AI is of particular relevance to tech-heavy sectors, with fintech startups having arguably more to gain from AI than any other business.

For example, AI can be used to automate high-scale data analytics, provide customer services through a chatbot, and offer more reliable information faster than any human. ChatGPT can even produce code, something which is traditionally seen as requiring highly specialist skills.

On the face of it, AI is brilliant. But do we all really understand what it means to use AI and integrate it into our business models? We doubt most people (consumers or businesses) have even read the ChatGPT Terms of Use before using it or its outputs…

So, what are the key risks involved in using or incorporating AI into your fintech business?

1. Output Ownership
One of the more pressing issues for fintech firms using AI outputs (particularly code) to support their business model is whether or not they own and have the right to use that output.

There is a risk that fintechs start integrating AI content into their infrastructure and relying on this heavily without the right to do so and are later served with a ‘cease and desist’ letter demanding payment for past use of someone else’s intellectual property. This could be a crippling cost for a small startup.

So, who actually owns the AI output? The default position in the UK is that a computer cannot own copyright. Instead, the copyright owner would be “the person by whom the arrangements necessary for the creation of the work are undertaken” (Article 9 of the
Copyright Designs and Patents Act 1988), meaning the person who owns the AI platform.

Therefore, unless the AI-owner actively assigns the outputs to you as the user of the AI (which many will be relieved to hear is the case in the ChatGPT Terms of Use), you will not own your outputs.

Top tip #1: always check the T&Cs of the AI platform before integrating its outputs into your systems.

That said, there is still the risk that the output provided to you has been wholesale ‘copy and pasted’ from another source and is already owned by someone else.

This is an issue which has gained a lot of attention in recent months, particularly given there is not always full transparency over what information is going in to AI or how AI learns its information.

Currently, there is an exception under the Copyright, Designs and Patents Act 1988 which allows AI owners to use any copyrighted work for AI machine learning, so long as the AI is only for “non-commercial research use”.

However, the Intellectual Property Office has recently proposed introducing a new exception to allow copyrighted work to be used for machine learning for “any purpose”, much to the concern of copyright owners across the UK and even the UK Culture, Media and Sport Committee.

The result is that AI outputs could contain copyrighted information (including yours as a fintech owner) which will put you at risk
of an infringement claim if you use it.

Ultimately, AI is much like a meat processor: if the data going in is of poor quality, so will be the data coming out. As a result, outputs could be unusable or inaccurate, and AI platforms often avoid responsibility for the accuracy of its outputs.

Even the ChatGPT Terms of Use note that “You should evaluate the accuracy of any Output as appropriate for your use case, including by using human review of the Output.” But if a human is required to validate every AI output, does this not circumvent the purpose of AI altogether, or even render it meaningless?

Top tip #2: always check that the output does not infringe any third-party rights before distributing this or using this in your business.

A related risk which we have alluded to above concerns the intellectual property which a fintech does own, whether that originated from Artificial Intelligence or plain old Human Intelligence.

How can a fintech business guarantee that AI won’t widely distribute this property to others? The short answer is that you probably can’t, and this is just a risk you should be aware of.

Interestingly, the Terms of Use for DALL-E2 (an AI system which allows users to create images and art from a description) expressly acknowledges this risk.

It states that “Due to the nature of machine learning, Output may not be unique across users and the services may generate the same or similar output for OpenAI or a third party… Other users may also ask similar questions and receive the same response. Responses that are requested by and generated for other users are not considered your Content.”

This begs the question: whose Content is this?

Of course, it is also possible that another human could independently generate the same intellectual property as you, but this is probably much less likely than the same AI system producing the same output multiple times over to different users. And then, going back to the ownership question, who would own this content?

If the ChatGPT Terms of Use say that you own the outputs generated for you, but if that output is also generated for and assigned to
your competitor, how does that work?

Is ownership purely based on a first-come, first-served basis? Ultimately, these issues still exist in the human world, but they are heightened in the context of highly evolved AI which operates at speeds and capabilities arguably beyond that of a human.

2. Regulation
Another consideration to keep in mind (albeit one which has not quite come to fruition yet) is the regulation which you will likely have to comply with when using or developing AI.

All the issues mentioned in this article have evidently provided governments cause for concern, as shown by the fact that global leaders have instigated an international summit on the safe use of AI (to be hosted in the UK in November 2023).

Governments across the globe are concerned about the potential legal and social impact of AI, with many governments (including in the UK) moving towards regulating AI.

While we know that regulators are generally slow to keep up with technology (and we are still playing catch up with internet regulation), this seems to be something which they want to get right.

Indeed, the European Commission has already published a proposed AI Act. In addition, the UK government published a White Paper on 29 March 2023 on AI regulation which recognised the need for regulation, albeit one which takes a “pro-innovation approach” and follows a more flexible, principles-based structure.

The UK government’s proposed approach would regulate the use of AI rather than the technology itself.

Top tip #3: keep alert to new rules and regulation in your jurisdiction.

There is of course the argument that regulation creates heavy compliance costs for businesses, which could be prohibitive for smaller fintech startups and actually hinder innovation and AI development.

Recommended reading

• Google outlines AI-assisted makeover
• Ofcom investigates telecoms industry over contract price rises
• Steven Seagal Hit With £240,000 Fine for Cryptocurrency Promotion Scheme

UK Finance, an organisation representing UK banks, recently criticised the UK government’s proposals for creating “unnecessary rules for financial services”.

There is also a question whether the UK government wants to impose additional costs on businesses in a time of economic uncertainty and turmoil. We will need to wait and see what weight the UK government gives to these criticisms when producing its
final proposals.

3. The Ethical Angle
Finally, fintech firms may also want to consider the potential ethical consequences of integrating AI in its business. Theoretically, you could use AI to replace a whole host of roles and jobs, from standard secretarial tasks to highly specialised roles like coding, especially as AI becomes more sophisticated and generalist.

AI could even potentially takeover C-suite executive roles, making business decisions which are free from human error, external influences, and competing interests. However, you may want to consider what the reputational impact could be if you choose to contribute to mass redundancy in a time of recession and a cost-of-living crisis, for the benefit of cutting a few costs.

Likewise, replacing humans with AI could have a longer-term impact on your culture as a firm and your satisfaction rate among customers.

Top tip #4: consider the impact of AI on your firm’s reputation and culture.

Ultimately, a fintech will be worried about one thing: numbers. You need to make profit, and if using AI as a smaller startup is going to reduce your cost and increase your profit, you will likely accept a lot of these issues as a commercial business risk. However, you should still be aware of these issues, which could rear their head as you grow and as AI develops.

Using AI responsibly and sensibly, potentially in accordance with current guidelines and principles, such as the recently published “Global Principles on AI”, is probably the best way to future-proofing your business against these risks and future regulation.

Note: References in this article to the ChatGPT and DALL-E2 Terms of Use mean theversion of those documents available on 10 September 2023.