Do you need help & advice with AI & Automation?
Business leaders, listen up! A new wave of AI tools is hitting the market, and if you’re not careful, they could put your organisation in serious trouble. ChatGPT’s Atlas and Perplexity’s Comet web browsers might seem like productivity boosters, but they’re not ready for business use just yet. They’re vulnerable to something called ‘prompt injection’, which is basically a way for bad actors to trick the AI into doing things it shouldn’t.
Key Takeaways
- Don’t allow staff to install or use Atlas and Comet browsers at work.
- These tools are susceptible to prompt injection attacks.
- The security risks outweigh any short-term productivity gains.
- Wait for stronger security controls before considering these tools.
What is Prompt Injection?
Prompt injection is a security flaw where malicious code is hidden within a website or shared data. When an AI tool like Atlas or Comet interacts with this infected source, the hidden code can trick the AI into ignoring its original instructions and performing unintended actions. Think of it like a hidden command that hijacks the AI’s behaviour.
Why Are These Browsers Risky?
Right now, the security measures in these new AI browsers aren’t robust enough. This makes it relatively easy for someone with bad intentions to inject harmful commands. They could potentially make the AI reveal sensitive information, perform unauthorised actions, or even spread malware. While the idea of speeding up work is tempting, the current insecurity of these tools means the risk to your business is just too high.
What Should Businesses Do?
Until better security controls are in place, the advice is simple: do not let your staff install or use these AI browsers on company devices or networks. It might seem like a hassle, but it’s a necessary step to protect your organisation from potential data breaches and other cyber threats. The short-term convenience just isn’t worth the long-term security gamble.
Need Help with an AI Policy?
If you’re feeling overwhelmed by the rapid changes in AI and need a clear policy for your small business, there are resources available. A template can provide a good starting point for businesses with fewer than 50 employees. Getting a handle on AI usage and security is important, and having a policy in place is a smart move for any business leader wanting to stay ahead of the curve without taking unnecessary risks.
Frequently Asked Questions
Which AI tools should a construction company ban straight away?
You should ban any AI tool that staff can use to upload client details, site plans, contract information or financial data without approval. If the tool is not approved by your IT provider and does not protect your data properly, it should not be used in the business.
Why are unapproved AI tools a risk for construction firms?
Unapproved AI tools can expose sensitive project information, client data and commercial details. They can also create security gaps, lead to data loss and cause problems with your legal and contractual duties.
What should I do before allowing AI tools in my organisation?
Set a clear AI policy, approve only trusted tools and make sure staff know what they can and cannot enter into them. Your IT partner should check the security, data handling and access controls before anything is allowed into use.
