Palo Alto Networks will significantly expand the scope of the cybersecurity offerings it makes available on Google Cloud in the New Year while at the same time making greater use of cloud infrastructure and artificial intelligence (AI) technologies provided by Google across its portfolio.

Specifically, Palo Alto Networks will make its Prisma AIRS platform for securing AI workloads and environments on Google Cloud available, along with its VM-Series of firewalls.

Palo Alto Networks has also committed to making its secure access service edge (SASE) platform, dubbed Prisma SASE, available as well.

Dan O’Brien, president and COO of the Futurum Group, said this expansion of the relationship between the two companies brings together a leader in AI and cybersecurity at a time when organizations need to keep pace with a threat landscape that continues to rapidly evolve in the AI era.

It’s not clear to what degree Palo Alto Networks will rely on Google infrastructure, but the multi–billion dollar alliance does provide it with an opportunity to extend the reach of its portfolio to a Google Cloud platform that hosts services such as Vertex AI, a managed service for building and hosting AI applications and agents.

While interest in building and deploying AI agents is running high, many organizations are also starting to have a greater appreciation for the cybersecurity implications. For example, it’s relatively trivial to create a prompt injection attack that can be used to instruct an AI agent to export sensitive data to a malicious actor. Palo Alto Networks, following its acquisition of Protect AI earlier this year, has been extending its cybersecurity portfolio to thwart these and other types of cyberattacks aimed at AI applications and agents.

Unfortunately, it appears the pace of adoption of yet another emerging technology is outpacing the ability of cybersecurity teams to secure it. Existing cybersecurity tools and frameworks are not easily extensible to AI applications and agents, which requires cybersecurity teams to invest in additional unfamiliar tools and platforms to secure. The challenge is that the level of risk that AI agents represent is substantially greater. Once compromised, an AI agent can be used to potentially compromise an entire end-to-end workflow.

At this juncture, it’s not so much a question of whether AI applications and agents will be compromised so much as it is to what degree. There have already been multiple cybersecurity incidents involving AI tools, platforms, applications and agents. The issue that remains to be resolved now is what level of priority will be given to securing AI environments before a series of breaches ultimately forces the issue.

In the meantime, cybersecurity teams would be well advised to proactively investigate how AI technologies are being adopted within their organizations. There is a long history of emerging technology adoption occurring in the shadows that often only comes to light when there is a major incident. By then, however, it’s no longer a question of prevention so much as it is an exercise in damage control that with a little extra forethought could have been avoided.