Latest News

If you’ve been in business for a while, you might remember ‘servers’. If you don’t, then think of a digital filing cabinet. A server was a physical machine that provided the back-office service of storing the digital data of a business. And like a filing cabinet, a place for it had to be found on the premises. 

This ‘on-premises’ server is more or less extinct for most small and medium sized businesses these days. What happened? The cloud happened. The cloud meant that large companies (Microsoft, Google, Amazon Web Services) built gigantic server farms. And then they leased the storage space to other companies. And that spawned a new generation of business tools like Xero, MYOB and QuickBooks Online.

But when the cloud came out, there were huge concerns amongst business owners and managers about data. Would it be safe? Would it be secure? Who would be liable in the event of data loss if the data wasn’t stored ‘on premises’?

This wasn’t the first time these concerns had come up. The arrival of the internet had brought similar worries and doubts. 

And now generative AI is causing the same sorts of questions to be asked. 

This article is about giving you an overview of these applications, then go through potential pitfalls of these tools and how you can manage the risk / reward trade-off effectively to gain the benefits of this new tool without taking unnecessary risks with your business. 

What is Generative AI?

Generative AI, also known as generative artificial intelligence, is a technology that focuses on creating new content. After a lengthy period of ‘training’ on vast data sets, the AI models can begin to generate original and creative outputs, such as images, music, text, or even videos. The best of these are basically indistinguishable from human-created content.

What Are Some Examples Of How It Can Be Used?

For business managers and owners, all of that might seem interesting, but a little abstract. So here are just three examples of how generative AI can be used within the four walls of a business in almost any industry:

• In-House Marketing
• Customer Service
• Emails composition or automatic replies based on previous examples

Practical Cyber Security Considerations

As cyber security experts, we have spent a lot of time thinking about, and talking to clients about how to evaluate these tools for security. We’ve narrowed down the main three ‘headline’ points. 

A. Evaluate Data Privacy and Security

Technology can often move faster than regulation (think of Uber versus decades old taxi rules). But even though Generative AI is available for use, it doesn’t mean it has been ‘approved’ by any regulator or industry-specific body.

And because these tools can apply in industries as diverse as bulk shipping to jewellery retailing, we’re unlikely to get a uniform set of regulation. So, what we can advise is that key decision makers ask themselves the following questions when evaluating the characteristics of a potential tool:

1. Do I understand if the tool complies with the data protection laws my business has to comply with? How is data that I input into the tool treated? Is it confidential? Or is it able to be used for other applications (such as further training the model).

2. Do I understand whether the data I input into the tool is: 

1. Stored, and, if ‘yes’,
2. How is it encrypted? 

3. Am I permitted under the terms of service and contracts I have signed with clients and suppliers to input data that could identify them into a third-party tool like this one? Do I require further informed consent to do so?

B. Assess Transparency and Fairness

In the same vein, here are some key questions to assist you in evaluating the transparency and fairness of any ‘outputs’ provided by a generative AI tool:

4. Have I tested the ‘outputs’ of the tool using ‘test’ questions where I am confident of the answers to check its accuracy and fairness according to my parameters?

5. Do I understand the potential for hidden or implicit bias to exist in these tools (given they are early stage and still in development)?

6. Am I comfortable with being held accountable for decisions / outputs provided by an AI model if queried by my clients and employees (e.g. if an AI tool is used to screen job applications for an open role for suitability)?

C. Check Reliability and Performance

And here are some key questions regarding reliability and performance of generative AI tools:

7. Do I have a backup in place if this tool becomes unavailable, or will key business functions suffer if it is ever unavailable?

8. How much reliance will [insert job function here] have if we introduce [insert generative AI application here] into their business-as-usual work processes? What percentage of the job will require the use of this tool?

9. Do we have criteria in place to measure the reliability and performance of the tool we are proposing to adopt. If not, what might those criteria be, and when will they be reviewed?

Key Takeaway

Generative AI tools like ChatGPT reached 100 million users faster than other milestone technologies like the internet or mobile phone. Chances are they are here to stay. But for businesses to adopt them effectively, care needs to be taken, especially with the data privacy and cyber security aspects. CyberUnlocked can partner with your business so that you can capture the benefits without compromising the security of your data or that of your customers and suppliers.