South African employees who use tools such as ChatGPT, Claude, Grok and Google Gemini at work could expose their employers to legal risks, lawyers have warned.
Law firm Bowmans highlighted potential conflicts between workplace AI use and South Africa’s existing laws, including the Protection of Personal Information Act (POPIA) and the Employment Equity Act. The country has no dedicated AI legislation, the report said.
AI decisions must allow for human input
Bowmans raised concerns about businesses using AI to screen CVs, conduct chatbot interviews or assess staff performance as reported by Briefly.co.za.
Under Section 71 of POPIA, decisions with significant effects on people cannot be based solely on automated data processing.
That means a job applicant or employee should be able to challenge a decision, understand how it was reached and engage with a human decision-maker.
This could apply when an AI system recommends rejecting an applicant or produces a performance rating that could affect someone’s promotion or dismissal.
Sharing personal information also carries risks
Staff who enter a colleague’s, client’s or another person’s personal information into public AI platforms may also create legal exposure for their employer, Bowmans warned. The firm said the employer remains responsible because it determines the purpose and means of processing the data.
The lawyers advised businesses to establish clear rules for using AI, restrict the information employees can share with such tools and train staff on their responsibilities under POPIA.
The warning highlights a challenge that workplaces in Kenya and elsewhere may also recognise: the convenience of AI tools must be weighed against the risks of sharing sensitive information.
Netizens Weigh In
Not everyone online was concerned about the growing use of AI. Commenters on MyBroadband’s Instagram page had plenty to say, with some defending the technology while others urged caution over privacy and sensitive information.
@kaygee_gp said:
“No one can stop AI!! I mean no one 🤣🤣🤣”
@owimaru_1 wrote:
“Everybody can get information, learn, and become whatever they source from AI for free. They are hating.”
@tsheporatshidi shared:
“It can even write a better affidavit than lawyers who think AI should be off-limits and learn to use it wisely.”
@mphoentle.legal noted:
“Posted about this months ago!!”
@deonventer offered a more measured take:
“Address is a POPIA data point, yet I find it with Google Maps. What people should watch out for is sharing company info with the chatbots.”
The debate comes as South African lawyers continue to warn about the privacy and legal risks associated with using AI tools at work, particularly when personal or confidential information is uploaded to public AI platforms.
