The interplay between GPU computing, cybercrime, and cybersecurity has been an increasingly evolving landscape. GPUs, thanks to their computational efficiency, have been repurposed by threat actors for activities ranging from hash cracking and cryptomining to the deployment of trojan implants underscoring a persistent eagerness to exploit GPU capabilities for illicit purposes. The advances and accessibility of the Generative AI technology this past year has proven that technological democratization goes both ways. It is time to leverage the robust power of GPUs more effectively in our defense strategies against cyberattacks.
In this session, we delve into Palo Alto Networks' approach to integrating Generative AI into our technology, while addressing the intricate security challenges it presents. We will cover the threat models that have shaped our co-pilot architectures, emphasizing our strategic framework decisions, such as the adoption of retrieval-augmented generation techniques. We will elaborate on the adaptations to the adversarial mindset while adjusting to this new paradigm and finally shed light on the AI/ML influence across our security technology portfolio, particularly in preparing for an AI-augmented threat landscape.