{
  "id": 12422248,
  "title": "The biggest AI security risk may already be inside your company",
  "url": "https://urgent.news/2026/10/06/the-biggest-ai-security-risk-may-already-be-inside-your-company",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-06T16:35:36.000Z",
  "source": {
    "name": "TechCabal",
    "slug": "techcabal",
    "url": "https://techcabal.com/2026/10/06/south-africans-bypassed-work-ai-rules/"
  },
  "original_language": "en",
  "account": "A recent survey reveals that half of South African employees have circumvented their company's AI restrictions in the past year, exposing businesses to a new security risk. This growing issue extends beyond traditional cyber controls, as employees now bypass corporate controls by copying company information into public AI tools, such as large language models (LLMs). Prejlin Naidoo, a partner at Oliver Wyman, shared these findings at a cybersecurity roundtable in Johannesburg. Naidoo emphasized that the corporate security perimeter now extends to employees and the AI tools they use. He explained that the Oliver Wyman Forum Global Consumer Survey 2026 South Africa report, based on responses from 903 South Africans, likely underestimates the issue, as the figure could be underreported.\n\nNaidoo pointed out that the risk may not be limited to junior employees, as managers are more likely to use unapproved AI tools due to their access to sensitive company information. This creates a bigger data-security problem than companies may assume. Even when employees have access to enterprise AI tools, the problem persists. Research from Kaspersky in 2025 found that 72.5% of South African professionals use AI tools for work, but only 30% received training on the cybersecurity risks associated with AI. Additionally, 35% of survey respondents had seen AI-generated work passed off as human work, raising concerns about the provenance and reliability of information entering corporate systems.\n\nThe survey also highlighted a growing trust in AI among consumers, with Oliver Wyman's research showing a fourfold increase in AI trust from 11% in 2023 to 44% now. However, 41% of those surveyed would be comfortable allowing an AI agent to place an order on their behalf. This growing trust expands ways AI can interact with company systems, customer information, and transactions, creating new opportunities for businesses. However, it also expands the potential points of vulnerability. At the Marsh cybersecurity roundtable, Spiros Fatouros, Marsh Africa and South Africa CEO, emphasized that cyber resilience must be viewed as an internal and external issue, considering the security of suppliers, partners, cloud providers, and other organizations connected to a company's operations.",
  "summary": "The corporate security perimeter now extends beyond networks and devices to employees and the AI tools they use.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}