Do you need help & advice with AI & Automation?
Business leaders, listen up! There’s a new development in AI that could put your organisation at serious risk if you’re not careful. Two new web browsers, Atlas from ChatGPT and Comet from Perplexity, have just been released. While they might seem like handy tools to speed up your team’s work, they’re currently not safe for business use.
Key Takeaways
- New AI Browsers Pose Security Risks: Atlas and Comet are vulnerable to prompt injection attacks.
- Prompt Injection Explained: Malicious code hidden in websites or shared data can trick the AI into performing unwanted or unsafe actions.
- Lack of Controls: Current security measures for these browsers are insufficient to prevent exploitation.
- Recommendation: Do not allow staff to install or use these browsers at work until better security is in place.
- Risk vs. Reward: The potential for a small productivity boost does not outweigh the significant security dangers.
The Danger of Prompt Injection
These new browsers, Atlas and Comet, are particularly vulnerable to something called prompt injection. If you’re not familiar with it, imagine this: someone can hide malicious code within a website or in a piece of data they share with the AI. This code tricks the AI into ignoring its original instructions and doing something else entirely – something the attacker wants it to do.
Right now, the security controls on these browsers aren’t strong enough to stop this. This means it’s quite easy for people with bad intentions to get these AI browsers to perform tasks you absolutely do not want them to do. It’s a bit like leaving your front door wide open – you wouldn’t do that at home, so why do it at work?
Why You Should Block Them For Now
On the surface, these tools might look like a great way to get your staff working faster. Who doesn’t want a productivity boost, right? But the reality is, they are too insecure for business environments at this moment. The risk of a security breach or data compromise is just too high.
Until proper security measures are developed and implemented, the best course of action is to not allow your staff to install or use these browsers on company devices or networks. It might feel like a hassle, but it’s a necessary step to protect your business.
Getting Help With AI Policies
If you’re feeling a bit overwhelmed by all this and need some guidance, there are resources available. For small businesses, especially those with fewer than 50 employees, there are templates that can help you create an AI policy. This can give you a starting point for managing how AI tools are used within your organisation.
If you’re unsure about setting up a policy or just need a hand figuring out the best way to use technology safely, don’t hesitate to reach out. The goal is to help businesses use technology effectively and securely, and that includes navigating the fast-changing world of AI.
