Page 203 - Cyber Defense eMagazine Annual RSA Edition for 2024

Traditionally, cybersecurity has relied heavily on manual intervention, making it challenging to detect and
respond to emerging threats in real-time. However, AI is revolutionizing this approach by leveraging
machine learning algorithms to analyze vast volumes of data and identify patterns indicative of malicious
activity. This includes detecting vulnerabilities in software, carrying out pattern recognition on large
amounts of data to recognise threats, scanning for malicious code or malware, and sending alerts in real
time.

In practice, if businesses want to use AI to detect or manage an attack, they can use synthetic generation
of datasets to provide examples which the AI can then learn from. For instance, one LLM can be used to
generate possible attacks, as many as we can create or imagine, and another LLM to learn from the
attacks. The second one can also be used as part of the defensive tools. Additionally, machine learning
can be helpful in discovering targeted attacks that were sitting within data sources and would not have
been discovered otherwise.

Hence, businesses must learn to leverage AI and threat modeling as quickly as possible otherwise they
may risk being targeted by threat actors. But it must be controlled and monitored carefully.

Integrating AI in business operations

Furthermore, beyond adopting security-by-design and threat modeling, and keeping on top of new
legislation, businesses should be cultivating a security-conscious culture among their teams across all
levels from C-suite executives to frontline employees. Everyone must understand the evolving cyber
threat landscape and their role in mitigating risks.

This can be in the form of regular training sessions to educate employees about the latest cyber threats,
including sophisticated AI-driven attacks, and how to recognise and respond to them. It also means
simulated cyber exercises and robust incident response protocols which can help bolster the
organization's resilience against cyber threats.

By encouraging a mindset where security awareness is part of day-to-day interactions within an
organization, instilling vigilance is crucial given that potential threats can emerge from seemingly benign
interactions.

Organizations should also invest in AI-ready infrastructure and talent capable of harnessing the full
potential of this technology. This entails recruiting data scientists, AI engineers, and cybersecurity experts
proficient at developing and deploying AI-driven security solutions tailored to the organization's specific
needs. It is going to be a collaborative effort but only then will we be able to get a handle on AI and the
threats it poses.
Looking externally, there will also be more collaboration between the cybersecurity industry,
governments, academia and civil society working together to come up with new ways to respond to
threats. The UK’s AI Safety Summit was a good example of how this collaboration can work in practice,
and we can expect this momentum to be maintained at the upcoming AI Safety Summits in South Korea
and France.

203