The long-term debate about AI is worth having. For a small business this year, five everyday risks matter more, and each has a simple fix.
- Date
- Reading
- 6 min read
- Written by
- Nick Li
- Tags
- #AI#Risk
Open the news this month and it is easy to believe the AI apocalypse is around the corner. The UN human rights chief has warned that the world is running out of time to put guardrails on AI, and a 35-member commission has listed AI among 17 "catastrophic threats" facing humanity.
Those conversations deserve to be taken seriously. But for a company of fifteen people, the more practical question is: how will AI affect my business this year?
The long-term debate and this year's decisions
Experts still disagree sharply about whether AI could one day slip out of control, and those questions sit with governments, research labs and the largest technology companies. What a small business can do is manage the risks it actually meets every day. These five deserve your time more than any doomsday scenario.
1. Staff pasting data into public AI tools
To save time, a colleague pastes a client list, a contract or a quote into a free chatbot. It is the most common risk and the easiest to miss. That data may be stored, and may even be used for training.
Safeguard: spell out which information must never go into public tools, and when sensitive data is involved, use services with business-grade privacy terms.
2. AI being confidently wrong
AI answers sound sure of themselves, but that does not make them right. If the assistant on your website quotes the wrong price or promises a policy you do not have, the responsibility is still yours.
Safeguard: have the AI answer only from information you provide, and keep a person reviewing anything involving prices, contracts or refunds.
3. More convincing scams
AI makes fake voices, fake videos and fake invoices remarkably convincing. In 2024, the Hong Kong office of a multinational company lost around HK$200 million after a video call in which every "senior executive" was a deepfake.
Safeguard: confirm every request to transfer money or change bank details through a second, known channel, no matter how much the caller sounds like the boss.
4. Betting everything on one tool
The AI tool that works well today may raise its prices, change its terms or shut down tomorrow. If your whole workflow is locked into one platform, switching later becomes very painful.
Safeguard: keep your own copies of data and process notes, choose services that let you export your data, and avoid putting your core business logic only on someone else's platform.
5. Doing nothing
The most underrated risk is not using AI at all. When competitors halve the time they spend answering enquiries and sorting documents, the gap widens quietly.
Safeguard: start with one repetitive, low-risk task, such as drafting replies or summarising meeting notes. Our article AI for small businesses: where to actually start has more detail.
A one-page AI policy
You do not need a thick policy document. A small team can adopt these five rules this week:
- Customer personal data and confidential documents never go into public AI tools.
- Anything AI writes for customers is checked by a person before it goes out.
- Any instruction involving money is confirmed through a second channel.
- Keep a list of the AI tools the company uses and who is responsible for each.
- Review once a quarter: what is working, and what should stop.
Rather than worrying whether AI will end the world, make sure it does not cause trouble inside your company.
If you would like help setting up a sensible way for your team to use AI, let's talk.