AI regulation. What do you need to know as an entrepreneur? | AI in business #47

Understanding AI regulations. What is it and how does it change business?

Artificial intelligence is a set of technologies and methods that allow machines to mimic human cognitive abilities. AI systems can learn and solve problems in ways that are increasingly similar to humans. A key role is played by machine learning and deep learning, which enable algorithms to optimize themselves based on data.

The applications of AI are almost limitless:

• generating content – text, as ChatGPT and Bard do, or images in Midjourney and DALL-E-3,
• translations – very useful for international business,
• image and sound recognition – used, for example, when inspecting the work environment with cameras or call center call recordings,
• advanced data analysis from a variety of sources – from social media to sensors from industrial sensors.

This is a true revolution that is changing the face of many industries. AI allows companies to automate processes, better understand customers, or develop new products. However, technology also brings risks – hence the need for regulation.

AI regulations: key concepts and principles

When implementing AI solutions in your company, you need to be aware of current regulations related to the use of artificial intelligence. The most important of these are:

• copyright,
• data protection law,
• laws regarding liability for the operation of artificial intelligence used by companies.

Copyright law vs. artificial intelligence

Copyright law protects the personal and property rights of creators. In many countries, only man-made works can be covered by its protection. As a result, lawyers tend to interpret that works created solely by AI in an automated process are not subject to copyright, even though artificial intelligence is usually unable to create content without human collaboration.

In the business context, the regulatory concerns related to authorship are:

• using the works of living creators to teach artificial intelligence – this applies to companies that create their models or teach existing ones using publicly available data; indeed, all works – from painting and sculpture to branding to texts published on websites – are subject to copyright,
• protection of company materials created with artificial intelligence – here it is necessary to carefully read the rules and regulations of individual AI tools to find out what regulations they invoke. Typically, however, they transfer the rights to the work created in cooperation with AI to the person typing the clues, or prompts for artificial intelligence, leaving themselves free to use the created content for promotional or research purposes.

October 30, 2023, marked the first verdict in the famous copyright infringement lawsuit brought by three American artists Sarah Anderson, Kelly McKernan, and Karla Ortiz against the three companies behind the leading image generators Midjourney, Stable Diffusion, and Dream Up.

The artists wrote in their lawsuit that artificial intelligence models were taught on their works without their permission, and thus could generate works that copy their works, violating their copyrights. However, the court dismissed their claims in a 28-page ruling, stating, among other things, that in the US, copyrights must be registered to sue for their alleged infringement. The artists were unable to identify which of their works were included in a dataset of images downloaded from the Internet and used to train artificial intelligence.

Law on personal data protection

The Data Protection Law (RODO) regulates the processing of personal data, which is central to many AI applications. AI systems often collect and process huge amounts of data, thus challenging businesses to ensure their security. For entrepreneurs, this means not only securing the acquired information but also ensuring its transparent processing and accountability throughout the process. In practice, it requires the implementation of effective security mechanisms and appropriate data management procedures to ensure that privacy is protected and users’ rights are respected.

AI accountability

New regulations are currently being developed to facilitate claims for damages caused by AI systems. The key issue here is the question of civil liability – whether the manufacturer, the user, or the system itself is responsible for the actions of AI. Pending regulations may introduce such solutions as:

• compulsory insurance for AI manufacturers, or
• special compensation funds.
• Therefore, it is important for entrepreneurs to closely follow these changes and prepare for the new requirements and obligations.

How is AI changing the legal landscape for small businesses?

The AI revolution is an opportunity for small companies – thanks to new technologies they can compete with the big players. However, the implementation of advanced AI tools involves meeting strict legal standards. And this requires financial outlays and knowledge that small players cannot always afford.

That’s why smaller entrepreneurs need to familiarize themselves with current and planned regulations in advance – this will allow them to minimize legal risks and implement AI responsibly. It is worthwhile to use the services of lawyers specializing in new technologies.

Source: DALL-E 3, prompt: Marta M. Kania (https://www.linkedin.com/in/martamatyldakania/)

Examples of AI applications in business and their legal AI regulation

What are the legal dilemmas in implementing AI in practice? Here are examples:

• chatbots – use of ChatGPT in a company under a non-enterprise license involves accepting the terms and conditions of participation in a “scientific experiment” conducted by OpenAI; the conversations conducted can be reviewed and used to create data sets for teaching the next generation of artificial intelligence,
• voicebots – can collect personal data without the callers’ knowledge, which raises privacy concerns; for the data relates not only to the subject of the conversation but also, for example, to a recording of the person’s voice. A small sample of it is enough to “clone” it, i.e. create a copy using artificial intelligence, which can create natural-sounding recordings of any content that will sound like that person’s utterances,
• personalization – algorithms that collect and categorize data about users may violate the principle of equal treatment; there is also a risk that they may be used for manipulation, creating information bubbles based on precise personalization of the message; therefore, one of the basic principles of AI ethics, as well as future regulations on the use of artificial intelligence, will be transparency in the use of data – but how it will look in practice is unknown,
• autonomous cars – require new regulations on safety and liability for possible accidents involving driverless cars.

As you can see, seemingly unrelated technical issues imply compliance with numerous regulations. And these are only selected examples closest to today’s business reality.

Key legal challenges of implementing AI regulation in small businesses

What specifically can give owners of smaller companies implementing advanced AI systems sleepless nights? These are primarily:

• complex and changing regulations – in the rapidly changing landscape of AI tools, it is difficult to keep up with technological innovations, while the legal regulations concerning them are slow to emerge, so it remains to follow the regulations of the tools used and to consult a lawyer specializing in the law affected by the company’s area of operation,
• liability for AI mistakes – this is one of the primary issues for business owners in implementing the latest developments in artificial intelligence, the question is who will be liable for any damages in the event of a breach of customer data protection or plagiarism committed by a social media post generator,
• rights to AI-created works – how to protect intellectual property if it is created in collaboration with artificial intelligence so that the protection is effective within a given legal system? This is a very important question in the absence of regulations and practical ways to distinguish between human-generated and human-created content,
• unintentional discrimination – how to eliminate algorithmic biases that can lead, for example, to the creation of segments of “first” and “second” category customers according to categorization rules understood only by AI.

These are fundamental dilemmas facing companies implementing advanced solutions based on artificial intelligence.

A guide for entrepreneurs: How to ensure compliance with AI regulations?

Laws regulating the use of artificial intelligence are still being developed in many countries. By force, it will be possible to ensure compliance with AI regulations only after they take effect. In October 2023, the White House released the Blueprint for AI Bill of Rights for the United States, which includes five principles to protect citizens from harm caused by artificial intelligence. The principles address the safety and efficiency of systems, protection against algorithmic discrimination, data privacy, information and explanation, and alternative options. Its detailed wording can be found here: https://www.whitehouse.gov/ostp/ai-bill-of-rights/

The regulation governing artificial Intelligence in the United States establishes new standards for security and AI protections aimed at protecting Americans’ privacy, as well as promoting innovation and competition, as well as strengthening the U.S. position as a leader in the development of artificial intelligence internationally. Key actions include:

• establishing requirements for security testing and sharing of results by companies developing advanced AI systems,
• developing AI security standards, tools and tests,
• combating discrimination caused by algorithms,
• protecting consumers, patients, employees and students,
• ensuring responsible government use of AI.

The European Union, in turn, is working on the “AI Act,” a set of regulations covering every application of AI in the EU. It is intended to cover all tools and interest groups – states, businesses and individuals – and to divide AI systems into risk categories:

1. Systems that pose unauthorized risks – for example, based on subliminal techniques,
2. High-risk systems – such as biometric identification or critical infrastructure management,
3. Limited risk systems.

The responsibility for the proper operation of high-risk systems falls mainly on their suppliers, which can become quite a burden for companies providing such solutions. The full text of the “Artificial Intelligence Act” can be found here: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex%3A52021PC0206

So how do you minimize legal risks when implementing AI in your company? Here are some practical tips. Firstly:

• focus on the anticipated impacts of implementing AI systems to catch potential risks in advance,
• from the beginning, keep in mind the mechanisms for controlling and supervising the operation of the algorithms, such as by human or automated tools that monitor their behavior,
• check that you are using trusted AI technology providers, read regulations and updates,
• Keep a close eye on updating AI regulations and comply with them.
• When in doubt, seek advice from lawyers specializing in new technologies.

The future of AI regulation. What can entrepreneurs expect?

AI regulation is still in its infancy, but it is already apparent that we are moving toward greater control and oversight of the technology. An example is the EU’s “AI Act,” imposing strict standards on high-risk systems.

On the other hand, there are voices, for example from the UK, advocating looser regulations that will not stifle innovation. Perhaps the future will be a golden mean?

Regardless of the exact solutions, it seems certain that AI regulations will become increasingly stringent, so businesses should take them into account when implementing new technologies in their companies.

Summary

The revolution in artificial intelligence is a huge business opportunity. However, when introducing advanced AI solutions in a small business, one should be aware of numerous legal pitfalls. Existing and planned regulations can give many business owners a headache. Therefore, it is crucial to gain an early understanding of the legal aspects of AI, conduct thorough risk analyses and comply with all requirements. Then both businesses and customers will be able to take full advantage of the benefits of artificial intelligence.

If you like our content, join our busy bees community on Facebook, Twitter, LinkedIn, Instagram, YouTube, Pinterest, TikTok.