You built or deployed an AI solution. How to get compliant?

Artificial intelligence has recently been a hot topic, and many businesses have decided to implement it in one way or another. Upcoming regulation is, however, now becoming a growing pain to deal with.

You built or deployed an AI solution. How to get compliant?
Photo by Jezael Melgoza / Unsplash

No matter if you are a software or an HR company, you should check how the upcoming Artificial Intelligence Act (“AI Act”) will affect your business. AI Act is still in the proposal phase and its final wording has not been agreed on yet. You can read more about it here. However, there are several good reasons why you should prepare for it beforehand.

Where to start?

The draft of the AI Act proposes a risk-based approach, meaning that different obligations will be posed based on whether you are a low-risk, high-risk company or use prohibited AI practices.

We are aware, that determining the appropriate category for your product or service can be challenging. As there are many borderline situations and the consequences could result in additional obligations (such as preparation of technical documentation, anonymisation of data, etc.), it is crucial to have your service or product correctly categorised.

No matter if you are planning to integrate AI-based solutions into your business, or if you already have, we can analyse your product, advise you on how high-risk you are, and what are your further obligations. Obligations imposed on you by the AI Act will also not be necessarily easy to implement. Dealing with regulatory matters yourselves will only take your focus away from where it should be – growth. We can take that burden for you and help you get done, everything related to complying with the rules set forth by the AI Act.

If your product falls within the high-risk category, we can guide you on how to adapt it in a way to lower your compliance obligations while keeping your product-market-fit.

It is not just the EU AI Act 

In addition to the AI Act, there are other fields of law, which are most likely relevant to your AI-powered solution. Here is a brief review of what awaits you.  

  1. GDPR/Privacy  

Who has not crossed paths with the GDPR yet? In the case of AI-based solutions, the challenges could be even greater. GDPR imposes various privacy protection measures on providers, and these measures are often even more difficult to implement in AI-powered solutions. Our “GDPR wizards” will help you find out how will AI Act apply to your business and what will be requirements if act as a processor or as a collector of personal data.  

  1. Terms of Use and Consumer protection  

Terms of Use are of huge importance for any business, as they determine ground rules. They also offer an important safety net for limiting liability, protecting content and preventing abuse. No matter if you are an HR or a software company, that uses AI, we can assess your product and advise you on how to adapt your policy in a way to suit your business.  

If you are still not convinced by the positive externalities, maybe liability for damages and fines proposed in the AI Act will make you change your mind. Administrative fines for non-compliance are high, up to EUR 20.000.000 or, if the offender is a company, up to 4 % of its total worldwide annual turnover for the preceding financial year.  

  1. Intellectual property  

Founders often focus on developing the best product possible, and along the way, they forget about the importance of intellectual property. This area already is, and is yet to be, quite problematic; what is more, it could be a huge step back for businesses.  

If you are a business owner, you are probably worried about which data was used to train AI models. In December 2023, the US newspaper, the New York Times, sued Microsoft and Open AI for copyright infringement, commercial exploitation and misappropriation of The Times’s intellectual property”. NY Times is demanding billions of dollars as the two companies have used articles, published by NY Times to train their chatbots. The answers to this lawsuit could set the tone for many future questions connected to intellectual property, but one can be certain that this is and will be an important topic in AI’s future development. Read more about it here if you are an AI geek:

The great regulatory battle between the EU and the USA – again. 

An everlasting question, when a new technology appears on the market, is whether to regulate it or not and, if yes, to what extent. The two diametrically opposite approaches could also be seen through the regulation of AI. EU has been a pioneer in the decision to regulate the market with the proposal of the AI Act. USA, on the other hand, decided for a looser approach, with a non-binding “AI Bill of Rights”. The EU seems to be able to provide more regulatory clarity but also imposes more restrictions at the same time. Time will tell, which will end up being friendlier to AI businesses.  

If you are not sure where and how to start, we are happy to discuss with you the benefits and hurdles in each of the respective continents.  

One of the most valuable business lessons is to not leave things till the last minute, especially when talking about legal and compliance matters. The legal landscape of regulating AI has recently been under the effect of many lawsuits connected to the use of training data, resulting in business owners becoming more aware of compliance issues. It is expected that the list of intellectual property and AI-connected questions will be ongoing, going hand in hand with technological developments.