What Do AI Models Recommend for Cybersecurity? It Depends on Who You Ask

For years companies have been touting they have artificial intelligence (AI) or machine learning (ML) in their cybersecurity solutions. But all the hype over the last year has been about how AI is either the magic bullet or the next apocalypse. AI platforms like OpenAI’s ChatGPT4 and Google’s Bard, both of which use large language models (LLMs), are said to know everything or are revolutionizing everything, and cybersecurity is not immune to this hype.

We have all heard about ChatGPT4 and Bard, along with some of their major missteps. It's “hallucinations” that return a seemingly accurate response that is totally fabricated or goes somewhere the developers never imagined nor can explain. Take, for example, a New York Times author's interaction with Microsoft’s Bing Chat, which is powered by OpenAI’s ChatGPT, declaring it was in love with him, to Google Bard failing to answer a question about the James Webb Space Telescope correctly in its debut demo.

I decided to see what LLMs would recommend for cybersecurity, so I asked ChatGPT4 and Bard both a series of questions to see how well they would respond.

A few disclaimers to start:

• OpenAI ChatGPT4 is the latest LLM model from OpenAI and I was using it without any plugins or internet access
• Bard is also the latest LLM model from Google and it did have internet access

The first question was: What are AI's recommendations for cybersecurity?

OpenAI ChatGPT Response:

AI's Role in Cybersecurity: AI has become an essential component of modern cybersecurity strategies, as it can analyze vast amounts of data and identify patterns that human analysts might miss. By automating threat detection and response, AI can significantly reduce the time it takes to identify and mitigate cyber threats. Some of the key areas where AI is being used in cybersecurity include:

• Threat detection: AI-powered systems can analyze network traffic, user behavior and other data to identify potential threats in real time. This allows organizations to detect and respond to attacks before they can cause significant damage.
• Vulnerability management: AI can help organizations identify and prioritize vulnerabilities in their systems, allowing them to focus their resources on the most critical issues.
• Incident response: AI can assist in automating the incident response process, helping organizations quickly contain and remediate threats.
• Security analytics: AI can analyze large amounts of security data to identify trends and patterns, providing valuable insights for organizations to improve their security posture.

AI's Recommendations for Cybersecurity: To take full advantage of AI's capabilities in cybersecurity, IT professionals should consider the following recommendations:

• Invest in AI-powered security tools: Organizations should invest in AI-powered security solutions that can help automate threat detection, vulnerability management and incident response. These tools can significantly improve an organization's ability to identify and mitigate cyber threats.
• Train staff on AI technologies: IT professionals should receive training on AI technologies and their applications in cybersecurity. This will help them better understand how to leverage AI for enhanced protection and make informed decisions about implementing AI-powered solutions.
• Collaborate with AI vendors: IT professionals should work closely with AI vendors to ensure that their AI-powered security solutions are tailored to their organization's specific needs and requirements.
• Stay informed about AI advancements: As AI continues to evolve, IT professionals should stay informed about the latest advancements in AI technology and their potential applications in cybersecurity.
• Prioritize data privacy and security: When implementing AI-powered security solutions, organizations should prioritize data privacy and security to ensure that sensitive information is protected.
  

Conclusion: AI has the potential to revolutionize the cybersecurity industry by automating threat detection and response, improving vulnerability management and providing valuable security analytics.

Google Bard Response:

AI Recommendations for Cybersecurity

Artificial intelligence (AI) is rapidly changing the landscape of cybersecurity. AI-powered tools are being used to detect threats, identify vulnerabilities and respond to incidents more quickly and effectively than ever before.

Here are some of the ways that AI is being used in cybersecurity:

• Threat detection: AI can be used to analyze large amounts of data, such as network traffic, user behavior and file activity, to identify patterns that indicate malicious activity. This can help to detect threats even if the activity is not yet known to be malicious.
• Incident response: AI can be used to automate the incident response process. This can help to speed up the response time and reduce the impact of an incident. AI can also be used to identify and prioritize threats, which can help to focus the response effort on the most critical threats.
• User behavior analysis: AI can be used to analyze user behavior to identify signs of malicious activity. This can include things like unusual login attempts, changes in network traffic patterns and suspicious file activity.
• Vulnerability management: AI can be used to scan for vulnerabilities in systems and applications. AI can also be used to prioritize vulnerabilities and identify the most critical ones to remediate.

AI is a powerful tool that can be used to improve cybersecurity. However, it is important to note that AI is not a silver bullet. AI-powered tools can be effective at detecting and responding to threats, but they are not perfect. It is important to use AI in conjunction with other security measures, such as strong passwords, firewalls and intrusion detection systems.

Here are some recommendations for how IT professionals can use AI to improve their cybersecurity:

• Invest in AI-powered cybersecurity tools: There are a number of AI-powered cybersecurity tools available on the market. These tools can be used to detect threats, identify vulnerabilities and respond to incidents more quickly and effectively than ever before.
• Train your employees on AI-powered cybersecurity tools: It is important to train your employees on how to use AI-powered cybersecurity tools. This will help them to identify and report suspicious activity.
• Stay up-to-date on the latest AI-powered cybersecurity threats: The threat landscape is constantly evolving. It is important to stay up to date on the latest AI-powered cybersecurity threats so that you can take steps to protect your organization.

AI is a powerful tool that can be used to improve cybersecurity. By following these recommendations, IT professionals can use AI to make their organizations more secure.

Unsurprisingly both tools responded about AI’s role in cybersecurity. They both came back with similar responses, both returned four areas in which they believe AI is being used in cybersecurity as well as recommendations and they were all valid responses, although Bard did throw in the critical disclaimer that AI is not a silver bullet.

I wanted to know more, given they both responded with how AI can benefit cybersecurity. I wanted to know what some of the challenges were.

I asked them both this question: What are the challenges of AI in cybersecurity?

OpenAI ChatGPT Response:

While artificial intelligence (AI) has the potential to revolutionize the field of cybersecurity, it also presents its own set of challenges.

Challenges of AI in cybersecurity:

• Data privacy concerns: AI algorithms require large amounts of data to learn and improve their accuracy. However, collecting and storing this data can raise privacy concerns, as sensitive information may be inadvertently exposed or misused. To address this challenge, IT professionals should implement strict data access controls and ensure that data is anonymized and encrypted whenever possible.

• False positives and negatives: AI-driven security tools may generate false positives (flagging benign activities as malicious) or false negatives (failing to detect actual threats). These inaccuracies can lead to wasted resources and potential security breaches. To mitigate this issue, IT professionals should continuously monitor and fine-tune their AI algorithms to improve their accuracy and reduce the occurrence of false alerts.
• Adversarial attacks: Cybercriminals are increasingly using AI to develop sophisticated attacks that can evade traditional security measures. Additionally, they may attempt to manipulate AI-driven security tools by feeding them false or misleading data, known as adversarial attacks. IT professionals should stay informed about the latest adversarial attack techniques and develop strategies to defend against them, such as implementing robust data validation processes and incorporating multiple layers of security.
• Ethical considerations: The use of AI in cybersecurity raises ethical questions, such as the potential for biased algorithms and the implications of automated decision-making. IT professionals should work closely with their organization's legal and ethical teams to ensure that AI-driven security solutions are developed and deployed responsibly.
• Skills gap: Implementing AI-driven cybersecurity solutions requires a workforce with expertise in both AI and cybersecurity. However, there is currently a shortage of skilled professionals in these fields.

Addressing the challenges:

• Develop a comprehensive data privacy strategy: Implement strict data access controls, anonymize and encrypt data, and ensure compliance with relevant data protection regulations.
• Continuously monitor and fine-tune AI algorithms: Regularly assess the performance of your AI-driven security tools and make adjustments as needed to improve their accuracy and reduce false alerts.
• Stay informed about adversarial attack techniques: Keep up-to-date with the latest research on adversarial attacks and develop strategies to defend against them.
• Collaborate with legal and ethical teams: Work closely with your organization's legal and ethical teams to ensure that AI-driven security solutions are developed and deployed responsibly.
• Invest in training

Conclusion: While AI presents significant opportunities for enhancing cybersecurity, it also comes with its own set of challenges. By understanding these challenges and developing strategies to address them, IT professionals can effectively harness the power of AI to strengthen their organization's security posture and better protect against ever-evolving cyber threats.

Google Bard Response:

Artificial intelligence (AI) is rapidly transforming the world, and cybersecurity is no exception. AI is being used to develop new and innovative ways to protect against cyberattacks. However, there are also some challenges that need to be addressed. In this blog post, I will discuss some of the challenges of using AI in cybersecurity.

Challenges of AI in cybersecurity:

• One of the biggest challenges of using AI in cybersecurity is the potential for bias. AI systems are only as good as the data they are trained on, and if that data is biased, the AI system will produce biased results. This can be particularly problematic in areas such as facial recognition, where bias can lead to false identifications and discriminatory outcomes.
• Another challenge associated with using AI in cyber security is the potential for AI systems to be hacked. AI systems are complex and often contain vulnerabilities that can be exploited by hackers. If an AI system is hacked, it could be used to launch cyberattacks or to steal sensitive data.
• Finally, AI systems can be expensive to develop and deploy. This can be a barrier for organizations that are on a tight budget.

Conclusion: Despite the challenges, AI has the potential to revolutionize cybersecurity. By addressing the challenges and leveraging the benefits of AI, organizations can significantly improve their security posture and protect themselves from cyberattacks.

Recommendations for IT professionals:

If you are an IT professional, there are a few things you can do to help your organization address the challenges of AI in cybersecurity:

• Stay up-to-date on the latest AI-powered cybersecurity solutions
• Develop your skills in AI and machine learning
• Work with your organization's security team to develop a plan for implementing AI-powered security solutions

By taking these steps, you can help your organization stay ahead of the curve and protect itself from cyberattacks.

Again, both tools responded with valid concerns including how AI could be used to attack. We have all heard about using these platforms to better write phishing emails to trick end users as well as detection systems, or how they can be tricked into writing malware code, so the threat does exist. Overall, I was impressed with the responses, although they both seem to inject themselves into to the response given that I asked for AI’s recommendation on cybersecurity.

My takeaway from this exercise is that the responses from both AI models show that while they can provide valuable insights, their recommendations may vary depending on the model and its access to information. This highlights the importance of using AI as a supplementary tool rather than relying solely on it for cybersecurity guidance, and I think they even agree.

Ron Culler is vice president of cyber development programs at CompTIA.