The machines are not coming for your job, yet…
There are those who argue that the rapid uptake of artificial intelligence-backed applications and large language models (LLMs) will result in a jobs bloodbath the likes of which the world has never seen, and there are those who deflect with somewhat uncertain declarations that humans will prevail.
The firing frenzy is just getting started
If you follow global hiring and firing announcements, you should find yourself siding with the former cohort. On 15 April 2026, Forbes.com led with 1000 job cuts at Snap, the parent of Snapchat. The firm’s leadership told staff they were saving money to ramp up AI use across several initiatives. The article offered up a long list of similar decisions by major United States (US) tech firms, including Oracle, which cut 20000-30000 jobs to offset heavy investment into building out its AI infrastructure.
Meta, the owner of Facebook and other heavyweight social media properties, confirmed in March that it was letting 700 people go, but commentary published on Reuters suggests the firm could reduce its 75000 global workforce by as much as 20% in the coming years. Another software firm, Atlassian, cut roughly 10% of its headcount to “self-fund further investment in AI” while Salesforce completed a two-step process, cutting just under 1000 staff in February 2026, just six months after AI was blamed for the clean-out of 4000 support staff.
Your writer has attended dozens of insurance and investment presentations where the impact of AI has come up for discussion. In most cases, presenters argue that AI will give humans an edge, and that humans who learn to work alongside the technology will be reasonably safe from redundancy. They paint a rosy picture of AI solutions like Microsoft’s Co-pilot pre-populating support staff email replies alongside AI-powered chat-bots handling complex queries online, both boosting productivity 10-fold. To play devil’s advocate, when have you ever heard of a private sector firm, answerable to its shareholders, retaining all of its staff in a 10x productivity world.
An insurer’s perspective
AI was top of the agenda at a recent media round table with Old Mutual Insure CEO, Soul Abraham, and Chief Actuary, Riccardo Govender. Early in the conversation, the South African short-term insurer hinted that legacy issues around insurance uptake, or penetration, could be tackled using AI. The technology is already being deployed to build insurance solutions that require less human intervention and are thus more affordable. “If you can make insurance 40% cheaper, then you could triple penetration,” Govender estimated.
Financial services firms are already using AI in areas like fraud detection, faster claims or onboarding processing and personalised customer experiences. “There are valuable gains from operational efficiencies in terms of making things faster and more automated,” Govender said. He mentioned that AI excels at complex scenario modelling in an interconnected risk environment. Firms cannot, however, rush into AI without weighing up a plethora of associated risks. These include hallucinations; overconfidence and incorrect outputs; data leakages and privacy issues; and cyber threat expansion.
In a recent thought leadership piece by law firm Webber Wentzel, Kim Rew, a partner at the firm, said that AI was radically changing the fraud landscape in both the combatting of fraud and perpetrating fraud contexts. “Criminals are now using AI to create synthetic identities, combining stolen real IDs with fake names and AI-generated images to bypass an insurer's onboarding verification,” she said.
Another major issue with rapid AI adoption is the inability of firms to look through the decision path taken by AI. “There are some challenges around the lack of explainability [and] the risk of models drifting over time,” Govender said, commenting on the black box that typically sits between data inputs and outcomes in this field.
The murky world AI decision making
Concerns over look-through capabilities were echoed in the Webber Wentzel article. “The Financial Sector Conduct Authority (FSCA) and Prudential Authority (PA) are urging financial institutions to adopt robust governance frameworks, ensure board-level oversight, and use recognised ‘explainability methods’ so that AI-driven decisions are transparent and auditable,” they wrote. “They also specifically mandate that institutions must clearly disclose when AI influences consumer-impacting decisions, such as credit assessments or insurance pricing.”
Returning to the media event, journalists wanted to know how AI adoption might impact headcount. “You can automate data cleaning; you can automate processes; but you cannot automate responsibility,” Govender said. “That responsibility lies with the human.” The Chief Actuary highlighted a longer-term risk of relying too heavily on AI and automation, saying that firms risked destroying the training ground of future human capability. To illustrate: how will a future employee take informed human decisions if they do not understand the processes?
“If the mundane work is being done by AI, we do need to fundamentally change how we work, and that will come down to employers investing in their people and trying to grow their people,” Abraham said, adding that AI would become pervasive in much the same way Google search has done, and that tools like Co-pilot were already becoming embedded organisationally. “I do not see AI replacing humans tomorrow; but the people that can get the most out of AI will grow in the insurance or any other business,” he said.
Major loss and liability claims loom
The risk to head counts is perhaps surpassed by potential losses or liability arising from AI use. Webber Wentzel warns that the rapid AI adoption has seen a growing list of incidents across industries. They cite Stanford University’s 2025 AI Index which revealed that AI-related ‘incidents’ reported worldwide in 2024 jumped by 56.4% from the previous year. “Any organisation using AI faces potential exposure,” wrote Rew. And the courts are already weighing in on the risk.
“In March 2026, a California jury found media platforms Meta and YouTube liable for USD3 million in a damages claim relating to their algorithms; Tesla was held liable for a fatal vehicle accident involving its autopilot system; and Air Canada was forced by a tribunal to honour a discount mistakenly promised by its chatbot,” the law firm wrote. “Closer to home, the FSCA has raised alarms about deepfake videos of prominent figures endorsing fraudulent schemes.”
Webber Wentzel also noted that lawmakers were scrambling to catch up. “The European Union has adopted its comprehensive AI Act, and Denmark is looking at copyright protection for individual likenesses against deepfakes,” they write. Meanwhile, in South Africa, a draft National AI Policy Framework was published in 2024 and is expected to be gazetted for a formal 60-day public consultation process soon. Finalisation is only anticipated during the 2026-2027 financial year.
AI in the financial sector
Leading financial services providers are in the firing line domestically. In November 2025, the FSCA and PA jointly published a landmark, first-of-its-kind report titled ‘Artificial Intelligence in the South African Financial Sector’ to offer a clearer picture of where the industry stands. “Banks are leading AI adoption at 52% while the insurance sector has adopted a markedly more cautious stance, adopting AI at just 8%,” Rew wrote. “However, insurers plan to expand their use of AI heavily into underwriting and claims management.”
Heavily, but still cautiously, if we adopt the tone from the aforementioned Old Mutual Insure event. Abraham believes the world is still in the first chapter of its longer-term ‘dance’ with AI. “We have not yet seen anything near the potential we could from AI,” he said. He advocated for a ‘go slow for now’ approach through which insurers could assess, adopt or abandon innovations at a modular level.
The first step for wider adoption is to lay a foundations to take advantage of AI, notably from a data stack and technology perspective. The second step involves taking advantage of large language models by integrating them across the data and tech foundation.
“Considering these profound changes in the way businesses are operating, the risk landscape has fundamentally shifted, making adequate insurance coverage essential,” concluded Rew. “Right now, most policies cover AI risks through silent cover, meaning AI is not explicitly mentioned, but the risks fall under the general policy wording.” She added that, as AI claims inevitably rise and coverage disputes develop, the insurance industry will move steadily toward clear, affirmative AI policies.
Balancing AI-human interactions
“The value from AI and data is coming out in underwriting and pricing and better risk selection,” concluded Abraham. “Much of the value is also coming from [AI-enabled shifts in] customer experience.” To deliver the desired customer experience, financial services businesses will need world class back-end systems and an appropriate balance of AI and human across client-facing functions.
Follow the writer on
LinkedIn: https://www.linkedin.com/in/gareth-stokes-media/
X: @stokesmedia
