The AI Risk Zimbabwe's Executives Are Not Naming

A Trevor & Associates Insight Report

Artificial intelligence (AI) is already running parts of Zimbabwe's private sector. Our survey of senior decision-makers found that almost nine in ten have deployed it in the past year. They can tell you what it threatens. They rate cybersecurity and data privacy as critical, and they are right to. But the risk that sits closest to them, inside their own buildings, is the one they rank last. And the single capability that would let them manage any of it is the one almost none of them are asking for.

At Trevor & Associates we put a questionnaire to Zimbabwean executives following our ThinkTank'26 convening in Harare. Thirty senior leaders began it. Seventeen completed it, across thirteen sectors, from micro firms of fewer than ten people to organisations employing more than two hundred. Their answers are more revealing than the headline adoption figure suggests, because the interesting findings are not in any single question. They are in the places where the answers contradict each other.

Adoption has outpaced readiness

Fifteen of the seventeen organisations surveyed have implemented AI tools in the past twelve months. This is not confined to innovation teams. The largest share of deployments is in workflow automation — scheduling, reporting and document processing — followed by customer support chatbots, HR and recruitment tools, and data analytics. These systems are embedded in ordinary daily operations.

Yet only seven of the seventeen rate their organisation's readiness to adopt and deploy AI as good or excellent. Four rate it poor outright. The remaining six describe themselves as average, which in the context of a technology reshaping decision-making is not a comfortable place to be.

That is the first contradiction, and it is the familiar one. What follows is less familiar and considerably more useful.

The constraint is capability, not capital

Ten of the seventeen executives named the lack of internal skills and expertise as their single greatest barrier to effective AI deployment. Financial constraints came second, cited by four. Infrastructure limitations, employee resistance and the absence of executive strategy were named once each.

The obvious reading is that Zimbabwean organisations cannot afford to do AI properly. The data does not support it. Self-rated readiness is distributed almost evenly across every organisation size band in the survey. Micro firms of under ten employees produced one of the three excellent ratings. So did a large organisation of over two hundred. Large organisations also produced a poor rating, as did medium and small ones. Whatever separates the ready from the unready in this sample, it is not headcount and it is not budget.

It is management.

What separates the ready from the rest

The three organisations that rated their readiness as excellent share a profile that no other group in the survey shares.

They are deploying more broadly. On average they have AI running across four distinct categories of use, from customer support to supply chain to decision support. Every other readiness band averages fewer than two. They all have formal AI policies either in force or actively in development. And every one of them has run staff training of some form in the past twelve months.

This matters because it inverts the intuitive sequence. The assumption in most boardrooms is that governance and training are what you build once the technology has settled. In this sample, the organisations that governed and trained early are the ones now confident enough to deploy widely. Preparation did not follow deployment. It enabled it.

Concern has not become control

Fourteen of the seventeen executives rate cybersecurity as extremely important in AI deployment. Thirteen say the same of data privacy. These are not leaders who are unaware of exposure.

Only four organisations in the entire survey have a formal policy governing AI use. Nine have nothing at all. Four more say a policy is in development.

Put those two findings side by side and the picture is stark. Of the fourteen executives who called cybersecurity extremely important, half have no AI policy of any kind. They have correctly identified a risk to their organisation and installed no mechanism to manage it. Staff in those organisations are using AI tools on live company data with no rules about what may be entered, what may be shared, what must be checked, and who is accountable when something goes wrong.

Risk awareness without risk management is not caution. It is negligence with a paper trail, and it is the position most organisations in this survey currently occupy.

The risk that ranks last

We asked executives to rate the importance of five considerations in AI deployment. The order they produced is worth reading carefully.

• Cybersecurity, rated extremely important by fourteen of seventeen

• Data privacy, by thirteen

• Transparency and explainability, by ten

• Regulatory compliance, by ten

• Employee impact and job displacement, by six

Employee impact ranks last, and by a wide margin. Every consideration ahead of it protects the institution: its systems, its data, its legal position, its defensibility. The one that protects the people inside it comes bottom of the list.

The breakdown by organisation size sharpens this further. Not one of the four large organisations in the survey rated employee impact as extremely important. Not one of the three small ones did either. The concern is real but unevenly held, and it is weakest precisely where the most jobs sit.

This is not a moral observation alone, though it is that. It is a commercial one. Workforce disruption that is not planned for arrives as attrition, resistance, industrial dispute and reputational damage. One respondent had already spotted it, warning that the biggest mistake they see executives make is treating AI as a technology project rather than a business transformation. A technology project has a budget and a vendor. A business transformation has people who need to be brought with you.

Confidence is built, not inherited

Employee confidence in using AI tools is low across the sample. Three organisations report extremely confident staff. Six report staff who are not confident. The rest sit somewhere in the middle.

The pattern underneath is clean. All three organisations reporting extremely confident staff had run AI training of some form in the past twelve months. Not one of the organisations that had run no training reported staff at that level of confidence. Most of the training reported was informal: only two organisations in the entire survey have run formal, structured programmes. Seven have run none at all.

So the picture across most of Zimbabwe's surveyed private sector is this: people are being asked to use consequential technology on real company data, without training, without policy, and without confidence. The organisations doing this are then rating their own readiness as average and expecting it to work out.

The advice executives give does not match the priorities they set

The most telling result in the survey is buried in the capability question. We asked executives which AI capabilities their organisations most urgently need, allowing up to two selections each.

AI strategy and leadership led, with nine mentions. General digital literacy and technical AI development skills followed with eight each. Data analytics and interpretation drew five.

Ethics and governance knowledge drew one mention from seventeen executives.

One. In a survey where thirteen of seventeen organisations have no AI policy in force, where half the leaders most worried about cybersecurity have no framework to act on that worry, and where the workforce consideration ranks last of five, governance is the capability that almost nobody has identified as urgent.

That gap is visible in the free-text answers too. Asked what advice they would give fellow Zimbabwean executives, respondents overwhelmingly wrote about people. Educate. Train everyone at every level. Do not rush in without preparation. Keep a human in the loop. Create space for teams to experiment. Confidentiality and ethics are key. When these leaders speak in their own words rather than choosing from a list, they describe exactly the capability their formal priorities leave out.

They know what is missing. It has simply not yet been named as a thing to buy, build or ask for.

Optimism is not a strategy

Despite all of this, eleven of the seventeen expect AI to deliver largely positive transformation over the next three years. Three more expect moderate improvement with manageable risks.

That optimism is not unreasonable. The productivity case for AI in Zimbabwean businesses is real, and the survey shows organisations already realising parts of it. But optimism held alongside no policy, no training and no workforce plan is not a forecast. It is a hope, and hope is not a control environment.

It is worth noting who the pessimists are. The two organisations expecting significant disruption and the one expecting minimal impact come from the public sector, financial services and law: the three settings in the sample with the highest accountability requirements and the least room to be wrong. They may simply be seeing the exposure earlier than everyone else.

How Trevor & Associates can help

The gaps this survey identifies are precisely the terrain Trevor & Associates was built to work on. We sit at the intersection of strategy, technology, communication and governance, which is where AI readiness is actually made. Whether your organisation is at the start of its AI journey or strengthening a deployment already running, we can help with:

• AI deployment strategy — translating ambition into a structured, phased and board-aligned roadmap

• Governance and policy frameworks — developing the internal policies, accountability structures and risk management protocols your organisation needs before the next incident rather than after it

• Digital and AI literacy programmes — equipping leadership teams and staff with the confidence and competence to work with these tools effectively

• Strategic communications — helping you explain your AI strategy credibly to customers, employees, investors and regulators

• ESG and ethical AI advisory — aligning deployment with your values commitments, stakeholder expectations and emerging global standards

• Stakeholder engagement and ethical lobbying — helping you shape the policy and regulatory environment in which you operate

AI is not a technology problem with a technology solution. It is a leadership challenge that demands strategic, communicative and ethical responses in equal measure.

If you are ready to move from awareness to action, we are ready to work with you. Contact Trevor & Associates at trevorandassociates.com

About the survey

The Trevor & Associates AI in the Zimbabwean Workplace survey was conducted by online questionnaire among executives and senior decision-makers, opening at the ThinkTank'26 convening in Harare in February 2026 and running over the three months that followed. To the best of our knowledge, it is the first executive-level survey of AI adoption and readiness in Zimbabwe's private sector conducted by an independent advisory firm, and Trevor & Associates intends to track these questions in future editions.

Thirty executives began the questionnaire and seventeen completed it, a completion rate of 57 per cent. All figures in this article are calculated on the seventeen completed responses and are reported as counts alongside proportions where the base is small. Respondents represented thirteen sectors including financial services, ICT, government, healthcare, manufacturing, aviation, law, media, education and professional services, and spanned micro, small, medium and large organisations. Responses were anonymous.