Legal Case Studies

Financial institutions in South Africa are using AI to detect fraud and prevent money laundering. AI algorithms can analyse large amounts of data and identify patterns, anomalies, and suspicious transactions that would be difficult for humans to detect.

Incorporating AI into businesses:

To effectively incorporate AI, businesses must first conduct a risk assessment to identify potential risks associated with their AI systems and ways they can mitigate such risks. Thereafter, it is important to develop a plan for incorporating AI into the business on a small scale, ensuring that the plan complies with the legislative framework and is not unethical.

Compliance:

AI compliance is still a grey area in South Africa and other countries as governments and businesses are trying to enact legislation that regulates AI. This has proven to be difficult due to the constantly evolving nature of technological advancements.

AI has come under attack in recent years because of lack of compliance and biases. In 2018, Amazon had to remove an AI covert which was said to be biased against women.

AI Compliance globally:

Compliance with AI includes different branches of law. For example, data protection laws, intellectual property laws, labour laws, and delict laws.

Legal compliance in South Africa:

One major important piece of legislation that has been enacted and is central to the use of data is the Protection of Personal Information Act of 2013 (POPIA). POPIA was enacted to promote the protection of personal information processed by public and private bodies and to introduce certain conditions to establish minimum requirements for the processing of personal information. POPIA does not explicitly state the regulation of AI, however, it is essential as it regulates the use of personal information.

Although not explicitly stated, AI systems that use personal information are also regulated in terms of POPIA, giving effect to section 14 of the Constitution (the right to privacy). Businesses need to be mindful of how they deal with data and make sure that they are not in contravention of POPIA, especially when it comes to the protection of personal data in terms of POPIA. Of major importance from the POPIA is how businesses control the data that is in their possession, and make sure that they store or dispose of such information for the purpose it was collected for.

Another legislation which courts may find relevant when dealing with AI cases is the Electronic Communications and Transactions Act (ECTA). ECTA is relevant to AI systems as it obligates businesses to ensure that AI systems are secure, and provides for the right of consumers to have access to their data in electronic form.

Ethical Considerations:

Furthermore, besides legal compliance, it is important for businesses to consider the ethical implications of adapting to the use of AI services. These ethical considerations include accountability, transparency, fairness, and privacy of data.

AI Body:

Another major development that has taken place is the establishment of the South African Artificial Intelligence Association (SAAIA) which will provide guidance on how to manage AI, assist in policy making at the provincial and national levels of government, improve research, and to make sure that AI is in line with ethics and regulations.

Lessons:

South Africa can learn from the European Union’s (EU) efforts to regulate AI. The EU is one of the leading organisations dealing with AI and has presented the Artificial Intelligence Act which is aimed at dealing with AI issues in a uniform manner. Brazil has also followed in the footsteps of regulating AI and have tabled a Bill in Congress.

However, as it stands, it is important for businesses to consult with experts before adopting AI systems due to the grey areas where AI is concerned.

However, the major hurdle that many businesses have dealt with is how employees conduct themselves on social media, whether they are on or off duty. The most used social media platforms include Facebook, Twitter, TikTok, and LinkedIn.

Although the use of social media has pros it also comes with cons. Some of these include employees posting controversial content on social media whilst in work uniforms, or in some instances in their private capacity, but it ends up having a negative impact on the employer’s reputation.

Case Law: In Juda Phonyogo Dagane v South African Police Services (2219/14) [2018] ZALCJHB an employee was dismissed for posting racist comments on Facebook. The court was of the view that since the employee was a member of the South African Police Services (SAPS) and there was no social media policy by SAPS, they concurred with the arbitration award by the Commissioner which was based on “common sense”. The court highlighted the importance of educating employees on the right to freedom of expression as it does not mean that they are allowed to say anything.

Way forward: Businesses around the world have incorporated social media policies to deal with issues of misconduct, especially by employees, and some of these policies are well detailed and thorough as they explain to employees what is required from them and also the consequences of their actions.

It is important for businesses to be well equipped on how to deal with social media, especially in relation to their employees. This can be done through training workshops for employees, a well-detailed social media policy, and mechanisms put in place to monitor and enforce the social media policy of the business.

ChatGPT and other similar systems using Artificial Intelligence (AI) have taken the world by storm, with the promise that it will streamline processes, reduce costs, and enhance efficiency. While AI is certainly capable of doing so, a recent judgement in the Regional Court for the Regional Division of Gauteng has indicated that the output of ChatGPT should be treated with caution and circumspection.

Parker v Forsyth N.O and Others: The Plaintiff, Parker, had sued the Defendants for defamation and iniuria due to certain alleged defamatory statements made against her. In this matter, the Court was required to decide on the Defendants’ exception to the Plaintiff’s particulars of claim, and application to strike-out certain allegations in the Plaintiff’s particulars of claim, both of which were granted, and the plaintiff’s claim was dismissed with costs.

In preparing for the hearing of this matter, the Plaintiff’s attorneys used ChatGPT for legal research, to identify case law relevant to their argument. ChatGPT provided a list of useful and relevant case law including full case citations and case summaries. The attorneys accepted this output and proceeded on the basis that ChatGPT was accurate. No verification of the output was conducted.

It was later discovered that not one of the cases identified by ChatGPT existed. Although not punitive in nature, the Court granted the Defendants’ requested cost order against the Plaintiff for 60% of the Defendants’ costs. This was, in part, because their failure to verify the ChatGPT output, and the subsequent investment of time by the Defendants’ attorneys to find the fictitious case law, made the cost order appropriate.

Not only was the plaintiff’s case dismissed and she was ordered to pay costs, but the attorneys have suffered tremendous embarrassment as a result of their blunder.

Lesson: Although this matter applies specifically to legal services, the lesson can be applied across professions. If your business makes use of AI, do not assume that its output is accurate and correct. While AI output is useful for a high-level starting point, apply your mind to the information provided by AI and verify its contents. Not doing so could have dire consequences for your business, such as penalties, breach of contract, and litigation.

AI remains a useful tool for businesses. It is up to you to make sure that you understand how to use it and acknowledge and compensate for its limitations, to manage and reduce risks for your business.

Although our Courts are not required to follow this decision, it may be used as a persuasive source if a similar matter came before them. It is therefore important that you keep this in mind as it may become relevant to you and your business in the future and may result in liability if you have been absent-minded in your use of emojis.

The Case: In this matter, South West Terminal Limited (SWT) sought summary judgment against Achter Land & Cattle Limited (Achter) in respect of the breach of a contract for the delivery of flax, allegedly concluded on 26 March 2021 (the Contract). SWT did not deliver the flax to Achter in terms of the Contract and denies that the contract was ever concluded between SWT and Achter.

The court considered two issues: the application of the Canadian Sale of Goods Act and what a thumbs-up emoji meant on the facts of this case. For our purposes, we will focus on the second issue.

SWT and Acher had a long-standing business relationship. Ordinarily, SWT would send Achter a message attaching a contract which stated “Please confirm the durum contract.” Achter would then respond with “Ok”, “Yup” or “looks good” and would proceed to perform in terms of the contract.

In the contentious flax Contract, a similar process was followed by SWT, who sent the Contract to Achter with the message “Please confirm flax contract.” Achter responded with a thumbs up emoji.

Achter argues that he was confirming receipt of the Contract rather than the contents of the Contract. SWT contends that, in keeping with the usual way that business was done, the emoji amounted to acceptance of the Contract.

The Court found that, on a balance of probabilities, Achter had approved the Contract using the emoji, as he had done before by saying “yup” or “ok”. The Court was also of the view that the emoji was sufficient to indicate agreement by Achter on the facts, which created a binding contract between Achter and SWT.

Accordingly, the Court granted summary judgment and damages against Achter.

The Lesson: As the use of digital means of communication advances, Courts will be required to acknowledge responses like emojis as creating binding contracts between parties. Always be certain that your response and behaviour does not relay an unintended idea to another party, particularly where agreement to a contract, and its resultant obligations, is involved. A Court, may, in this instance, find you liable for breach of contract.