Managing risk: How to prepare your organisation for the risks of everyday AI

Artificial Intelligence (AI) can do so much. It can reduce busy work, and uncover new ways of looking at our data, finding patterns and ideas we might not have thought of naturally. However, it’s important we’re keeping our data, our organisation’s data and our clients’ data safe before integrating or implementing new AI systems. Here are a few tips to help you manage the potential risks of implementing new tools.

Knowing what you need to know

It’s worth starting with a clear-eyed view of what most not-for-profit (NFP) AI adoption actually involves. The majority of organisations across the APAC region are not building custom AI models or setting up extensive machine learning analytics. They are instead, purchasing subscriptions, configuring off-the-shelf tools and enabling AI features within systems already in use.

Even a basic content generation tool introduces risks relating to accuracy, data privacy and over-reliance on AI-generated outputs. Those risks escalate considerably where the work involves vulnerable populations, sensitive personal data or decisions that affect access to services.

Registering your risks

The risk register template accompanying this article is designed to make risk management practical rather than theoretical. It is structured around nine risk categories developed specifically for the social sector: accuracy, fairness, privacy, security, legal and compliance, human oversight, vulnerable populations, operational continuity and reputational impact. For each AI tool that an organisation adopts, the register should capture what the tool does, what data it handles, who is affected in the event of failure and what controls are in place. Critically, it should also record a named internal owner for each identified risk; an individual within the organisation with the authority and responsibility to act when an issue arises.

Proportionality is essential

Not every AI tool justifies a comprehensive risk assessment involving stakeholder consultation and board-level approval. A team using an AI writing assistant for internal communications presents a fundamentally different risk profile to a chatbot interacting directly with people in crisis. A tiered approach is recommended: low-risk tools require basic checks and a register entry; tools that handle client data or influence service delivery require more rigorous assessment, structured piloting, and ongoing monitoring; tools that serve vulnerable populations directly require the full set of safeguards, including defined human oversight, escalation pathways, the ability to shut the system down rapidly, and independent review.

Tools change, risks evolve

Regular reviews should be embedded into operational practice. The risk register should be revisited quarterly for the first twelve months following any new AI deployment, and at minimum annually thereafter. Each review should assess whether the risks originally identified remain current, whether new risks have emerged and whether existing controls are functioning as intended. AI tools evolve rapidly, vendors update their features and policies, and organisational usage patterns frequently drift from what was originally planned.

Take control through effective risk management

Risk management might not be the most exciting part of AI adoption, but in many ways it’s the most important one. With the included tools and guidance, we can hopefully streamline some of the processes for you – and get you back to your important work!

Managing risks related to AI usage is a team effort, and involves a mixture of education, process and policy. If you think you’re ready to take the leap into implementing AI for your organisation, check out some of our other guides and resources below.

Is your organisation ready for AI? If you’d like to test your organisations for AI readiness, complete our AI Readiness Assessment Quiz. 

Ready to go? Take the AI Procurement Quiz to discover your organisations pathway. 

Looking for expert advice? If you’d like support scoping your potential AI solutions, reach out to set up a mentoring consult with our AI experts.