5 Simple Guidelines For Smart AI Usage
December 31, 2025
5 Simple Guidelines For Smart AI Usage
CAPE TOWN, SOUTH AFRICA, NOVEMBER 2025
AI can be a useful and effective tool when doing all kinds of work. Whether you utilize AI in education, in your professional capacity or in your daily decision making, it’s important to be mindful of the risks involved in using AI.
At SABEN, we take your data seriously, so we’ve put together a set of practical guidelines to help you use AI platforms without putting your data at risk. Protect yourself and your organisation from AI misuse, data leaks and misinformation by adopting these best practices*.
1. Keep Personal Information Private
Do not input personal or confidential information into an AI tool.
Avoid including any of the following in an AI prompt:
- Personal identification (ID / passport numbers)
- Names, emails, phone numbers
- Addresses or banking information
- Employee or customer records
- Confidential company data or project details 1
Anonymise data
If you need to test or process real-world scenarios using AI, remember to replace sensitive information with placeholders, e.g. ‘<person’s name>’ or ‘<banking details>’.
Never use AI to make decisions about people
Don’t ask AI questions about people by name for the purposes of hiring, reviewing their performance, or performing disciplinary actions. Always use a verified, auditable human process for these functions.
2. Remember The 2 General Principles of AI Usage
AI has no ethical or legal boundaries.
Without human judgement, AI tools have the capacity to create and spread harmful, discriminatory, and illegal content. It is in both your and your organisation’s best interests to find and eliminate problematic content before publishing AI-generated content. Review everything you create and follow proper approval processes before you share it with a wider audience.
AI is a tool, not an authority.
AI software draws from legitimate and illegitimate sources alike, so it’s wise to treat AI-generated content as a draft, suggestion, or starting point – not as fact or policy.3. Review Content Responsibly
Always read over and verify AI output.
AI content can be incorrect, biased, or misleading. AI can also misjudge tone, cultural nuance or context. Check the facts, calculations, citations, and appropriateness of the content before using or sharing it.
Take accountability for the content you use.
When you make use of AI, always apply your own judgment and personal / organisational context to what the AI tool produces. The accountability lies with you to ensure that your final product is factual and contextually accurate.4. Respect Your Organisation’s Protocols
Use authorised AI platforms only.
Only use tools that have been vetted and approved by your organisation’s IT or data security team. At SABEN, we trust the following platforms:
ChatGPT: https://chatgpt.com/
Copilot: https://copilot.microsoft.com/
Do not connect AI tools directly to internal systems.
Avoid using plugins or APIs that access internal databases unless they have been specifically approved and secured by your organisation’s IT or data security team.
Log and track AI usage in critical business processes.
If you use AI to support decisions in customer service, marketing or risk analysis, ensure that you are transparent about your AI usage and that you record how you have used the AI for future audits.
Monitor the safety, efficiency and efficacy of your AI usage by reviewing the processes you use and seeing how these processes could run smoother.
5. Ensure Compliance
Make sure your AI usage is in agreement with:
- POPIA and other relevant data protection laws
- Internal data governance policies
- Copyright and IP guidelines
In Summary
✅ Stick to approved AI tools
✅ Use AI for support, not decision making
✅ Don’t enter personal or confidential information
✅ Follow privacy, ethics, and compliance rules
✅ Always review output before using
Possible uses for AI | Is this a sensible use of AI? | Comments |
| Drafting emails and content | ✅ Yes | Always review before sending |
| Analysing customer feedback | ✅ Yes | Use anonymised input. |
| Writing legal contracts | ⚠️Caution | Use as a draft only; review with a legal expert |
| Processing HR data | ❌ No | Never include identifiable employee data |
| Making hiring decisions | ❌ No | This must remain a human-led process |
Further reading: AI, authorship and authority by the University of Cape Town
*These are the guidelines we follow at SABEN. Please be aware that this blog post is intended to be educational, and is not legal advice. Please follow the best practices highlighted by your organisation and legal authorities.