Microsoft has developed a new chatbot, one designed to help cybersecurity professionals understand critical issues and to guide users with ways to fix various security concerns. Does this represent a good idea? Called the Microsoft Security Copilot, the system draws on GPT-4, the latest large language model from OpenAI.
As with any introduction of new technology, especially to the security sector, there are different opinions and the detail will full emerge once security personnel have been able to check out the system.
Nonetheless, some experts expressed concerns about this proto-AI development, not least because as chatbots will gain access to company IT systems. An example of a sceptic is Ivan Ostojic, who is the Chief Business Officer, Infobip.
Ostojic has told Digital Journal about the risks of using chatbots for security measures. He also assesses the types of industries for which this form of chatbot technology could prove the most beneficial.
Ostojic begins by setting out his central concerns: “Giving data access to a chatbot is risky, as it leads to security vulnerabilities and easier access to company data assets worldwide.”
This is not least because there is no going back, as Ostojic explains: “Once provided, that type of access is hard to retract, and can open the gateway for a myriad of much deeper security issues.”
Expanding upon this, Ostojic adds: “Data aggregation can also be unreliable in the case of cybersecurity, where information is being pulled from various sources and may not have a picture of the exact context of the security issue being presented.”
In terms of real-life examples, Ostojic notes: “As we’ve learned with ChatGPT, the information pulled can be useful, but may not have all components of the issue at hand to provide the best answer, rather than the right answer.” Errors can relate to factors like incorrect input format, API memory limit, input size, unstable internet connection, and other technical issues.
Looking at chatbots more generally, Ostojic assesses: “Chatbots have seen the most significant adoption in industries such as eCommerce, transportation, hospitality, and marketing, even before the introduction of ChatGPT.”
As to where this type of technology can make a difference, Ostojic cites: “Chatbots conducting customer service now have the sophistication and intelligence to talk like a human, responding quickly and efficiently. Rule-based and AI-based chatbots are being integrated into one area where we’ll see the most impact, as chatbots can now handle complex transactions such as returns and refunds.”
Therefore, a careful assessment is required for each sector primary to adopting these newer forms of technology. Here Ostojic recommends: “Across all industries, we must assess whether a chatbot will disrupt or complement what a human can do. Chatbots can reduce the human workload but will also need to be looked at closely for accuracy and efficiency. This technology has the potential to save companies significant costs of hiring. However, as chatbots evolve, the application and risks they bring to the industry must be evaluated.”
