Half of surveyed South Africans bypassed workplace AI restrictions, study finds

Half of the South Africans surveyed for an Oliver Wyman report said they had bypassed workplace restrictions on artificial intelligence tools during the previous year, highlighting growing concerns over data security and the reliability of work produced with such systems.

Prejlin Naidoo, a partner at Oliver Wyman, presented the findings at a Marsh cybersecurity roundtable in Johannesburg. The Oliver Wyman Forum Global Consumer Survey 2026 South Africa report was based on responses from 903 South Africans and was due to be released in November.

Naidoo said the reported 50% rate was likely an underestimate. She also said managers were more likely than their teams to use artificial intelligence in the unauthorised manner covered by the survey.

“This creates an entirely new kind of threat perimeter,” Naidoo said.

Data and workplace integrity concerns

Employees can move company information beyond corporate controls by copying it into public artificial intelligence tools, including large language models. Naidoo cautioned that simply providing approved enterprise products, or blocking unauthorised alternatives, may not be enough to control how employees use the technology.

The discussion also examined whether information entering company systems can be trusted. Naidoo said 35% of those surveyed had seen work produced by artificial intelligence presented as if it had been created by a person.

That trend raises questions about the source and dependability of information used in organisations. “Those are two really important areas that I think we need to watch,” Naidoo said, referring to the sharing of sensitive company information and the integrity of machine-generated work.

Survey results cited at the event indicated that 31% of employees in the communications, media and technology sector had used unapproved artificial intelligence tools, compared with 34% across other industries.

A separate 2025 Kaspersky study cited during the discussion found that 72.5% of professionals surveyed in South Africa used artificial intelligence tools for work, while only 30% had received training on the associated cybersecurity risks.

Risks extend beyond company networks

Spiros Fatouros, chief executive officer of Marsh Africa and South Africa, said organisations must consider threats arising from their wider business connections.

He pointed to recent cyber incidents involving South African insurance companies, saying attackers were able to gain access through suppliers and partners rather than directly targeting the affected companies’ systems.

“The key message is that cyber resilience can no longer be viewed solely as an internal IT issue,” Fatouros said.

He said suppliers, partners, cloud providers and other connected organisations should be part of companies’ approach to cyber resilience.

The event also reflected growing consumer acceptance of artificial intelligence. Naidoo said Oliver Wyman research found reported consumer trust had risen from 11% in 2023 to 44% at the time of the research. In addition, 41% of respondents said they would be comfortable allowing an artificial intelligence agent to place an order on their behalf.

Changing workplace roles

Keletjo Chiloane, a career and workforce consulting executive at Marsh, said the technology was changing more than company systems. It was also altering how work was organised and the skills employees needed.

“AI is not just a technology implementation. It changes how work gets done, which means organisations need to rethink roles, responsibilities and the skills people need to work alongside these systems,” Chiloane said.

The discussion therefore placed artificial intelligence use within a broader question of organisational preparedness, combining employee training, data protection, supply-chain security and the redesign of work between people and machines.