Artificial Intelligence in Business: Legal & Ethical Considerations

Artificial intelligence (AI) is an area of computer science where intelligent machines and software systems are designed to mimic human behaviour. AI solutions can be created for the healthcare, transport, financial and business, as well as manufacturing and retail sectors.

This article addresses some legal issues and ethical implications for business owners who intend to develop AI products or join the AI sector. It will cover:

• Whether a patent should be obtained for your AI invention
• Whether an AI invention is protected by copyright
• What happens if an AI causes potential harm to a human?
• What happens if an AI invention causes a data breach?
• Ethical considerations when creating AI solutions

What are Some Legal Considerations For Your Business’ AI Development?

Obtaining a patent for your AI invention

Is a particular AI invention likely to be made publicly available? Or could it be easily figured out by others? If so, business owners should protect their AI invention by applying for a patent.

Patents are monopoly rights given to an applicant for his patentable invention, which can be either a product or a process. This means no one can use, copy or make the invention unless the applicant allows it. The protection lasts for 20 years from the date of filing.

An AI invention is patentable if it meets all three of these requirements:

1. It is new. An invention is new if the global public does not know of it in any way. The invention should therefore be kept confidential before a patent is filed for it. In this regard, an organisation should consider preparing non-disclosure agreements and encrypting documents that mention the AI invention.
2. It has an inventive step. There must be an improvement over products and processes that are already available.
3. It is capable of industrial application. This means the AI invention can be made or used for practical purposes in any industry.

For more information, refer to our article on the requirements for patent registration in Singapore.

Generally, a patent application takes 2 to 4 years to be processed. However, IPOS offers an accelerated patent protection scheme called “SG IP Fast Track”, which is available until 29 April 2022. The scheme applies to patent applications in all technology fields and patent applications can be granted in as fast as 6 months. To be eligible for this scheme, your patent application must be first filed in Singapore.

Is your AI invention protected by copyright?

Business owners may have their AI inventions protected by copyright as long as they meet the relevant requirements. Copyright protects the expression of ideas but not the ideas themselves. Literary works and technological creations like source codes and data compilations can be protected by copyright.

The three requirements for copyright to subsist in the source code or data compilation for an AI invention in Singapore are:

1. The work is original. It should originate from the author and it must not be copied from another work.
2. It has sufficient connection to Singapore. The author is a Singapore citizen or resident in Singapore. Alternatively, the work is published in Singapore.
3. It is fixed on a tangible medium. For instance, the work is in writing on a piece of paper.

Registration is not required for copyright protection to be effective.

What happens if an AI causes potential harm to a human?

Each time an AI solution is invented, there will be benefits and risks. On the one hand, self-driving cars could reduce travel time. On the other hand, self-driving cars could also potentially malfunction and cause road accidents. Who is liable in such scenarios depends on the level of human involvement in the AI invention.

If a self-driving car causes an accident, the car obviously cannot be sued and there might not be any human driver in question. Hence, there is a need to examine the causes of the accident. If the accident had occurred due to the car’s features malfunctioning, or there had been a human driver in the car who failed to intervene to prevent the accident, then liability may be attributed to either the manufacturer or the driver respectively.

The Personal Data Protection Commission (PDPC) has rolled out a Model AI Governance Framework, which contains comprehensive recommendations that organisations can implement when using or developing AI solutions to reduce the risks of potential harm caused by AI solutions.

One such recommendation is a design framework that helps organisations regulate the level of human involvement in AI-facilitated decision-making. The design framework has two axes:

1. The probability of harm
2. The severity of harm to an end-user because of a decision made by the organisation that involved the use of AI.

Although the Model AI Governance Framework is not legally binding, the PDPC encourages organisations to adopt and adapt it for their own use. To mitigate risks, organisations should also carry out regular reviews of their AI inventions and test them in real-life situations.

What happens if an AI invention causes a data breach?

Another potential risk of using AI solutions is the possibility of the AI invention causing a data breach. A data breach could arise where, for example, personal data processed by the AI invention is leaked due to the negligence of the inventors, any person or employee involved.

In Singapore, data breaches are governed by the Personal Data Protection Act (PDPA). If there is a breach of the PDPA, the PDPC may impose a financial penalty of:

• For individuals: 5% of the person’s annual turnover in Singapore for persons with annual local turnover exceeding S$20 million, or up to S$200,000, whichever is higher.
• For organisations: 10% of the organisation’s annual turnover in Singapore for organisations with annual local turnover exceeding S$10 million, or up to S$1 million, whichever is higher.

This is separate from any fine or imprisonment that the court may impose on conviction for a data breach offence.

To manage and monitor the data collected by their AI solutions, business owners and employees involved should regularly review the AI systems or logs to identify possible security issues. Organisations should also implement privacy policies and practices to inform employees of their rights and responsibilities when managing personal data, and ensure that the public is aware of how the organisation uses the data collected or manages risks.

This could involve developing a data breach management plan to help identify potential data breaches arising from the use of the AI invention, and state the actions to be taken during such an event.

What are Some Ethical Considerations When Creating AI Solutions?

Presently, AI is a growing phenomenon and will be ethical concerns that organisations should address when deploying AI solutions. Only then would the public better trust and rely on AI inventions. In this regard, the Model AI Governance Framework highlights certain AI ethical principles for organisations to consider adopting, such as:

1. AI should not discriminate or violate internationally recognised human rights such as a person’s right to liberty.
2. Organisations should be able to explain the decision-making process within the AI invention in non-technical terms to people. This allows the public to probe, understand and trust the AI solutions.
3. The AI invention should be accurate and reviewable by third-parties, which can be done by creating and keeping audit trails and logs. This ensures that the operations are verifiable and transparent.

All in all, AI solutions must protect the interests of human beings, including their well-being and safety. AI actors, such as the designers, manufacturers, owners and operators of the AI invention, should also be held responsible for operating the AI system properly. This protects the public from potential harm. If harm is caused, individuals can seek legal redress from these AI actors.  Finally, organisations should keep records to identify the person(s) legally responsible for the AI solutions.

If you or your organisation wish to venture into the AI sector, you should seek a lawyer’s assistance in helping you decide the most appropriate protection for your AI invention. Furthermore, they can guide you on how best to address the potential legal and ethical issues that you may encounter.

You may get in touch with experienced technology lawyers here.