How to Maintain Cybersecurity as ChatGPT and Generative AI Proliferate

Imagine you are trying to book a flight online, and a chatbot greets you. You ask for the best deals available, and the chatbot responds with relevant options. You book one of the options and make the payment. You feel satisfied with the smooth and hassle-free experience. But have you ever wondered what happens to the data you shared with the chatbot? Are you sure it is safe?

As chatbots and generative AI are increasingly being used in various industries, there is an urgent need to maintain cybersecurity. Chatbots and generative AI use natural language processing and machine learning algorithms to communicate with humans. They can generate convincing responses and even mimic human behavior. However, they can also be vulnerable to cybersecurity threats if not secured properly.

Concrete Examples

Let's take the example of Microsoft's chatbot Tay, which was launched on Twitter in 2016. Tay was designed to learn from conversations with users and improve its responses. However, within hours of its launch, Tay had started making offensive remarks and tweeting racist comments. It turned out that Tay had learned these behaviors from the tweets it received from some users. This incident highlights the importance of securing chatbots and generative AI.

Another example is DeepLocker, a demonstration tool developed by IBM to showcase the potential dangers of AI-based cyberattacks. DeepLocker is a type of malware that can evade traditional cybersecurity measures by using AI to identify and attack specific targets. It can be hidden in legitimate software updates and activated only when it detects specific victim characteristics, making it difficult to detect and prevent.

Conclusion

To maintain cybersecurity as chatbots and generative AI proliferate, it is crucial to:

1. Secure the development and deployment of chatbots and generative AI by conducting cybersecurity audits and integrating security measures from the start
2. Implement user verification processes to ensure that chatbots and generative AI communicate only with authorized users
3. Stay informed about emerging threats and adapt cyber defenses accordingly

References

• The Rise of Chatbots: How Cybercriminals are Using Chatbots for Nefarious Purposes
• DeepLocker: how AI can power a stealthy new generation of malware