Last week we attended RSAC 2025 Conference in San Francisco. One of the biggest gatherings of security minds in the industry, with approximately 45,000 people attending. What were some of the main topics of the conference? We distilled them for you and discuss them in this brand-new Techzine Perspective.
A dominant theme at RSAC 2025 was the integration of AI, specifically agentic AI, into security tooling. While AI automation offers significant potential for streamlining security operations, the discussion emphasized the critical need for “human in the loop” oversight, particularly given the higher sensitivity and risk inherent in security compared to other sectors like call centers. This isn’t simply about automating tasks; it’s about building responsible and secure AI systems.
Subscribe to Techzine Perspective on YouTube
The conversation also touched upon the contrasting approaches of different vendors to AI automation in security, highlighting the industry-specific considerations and the higher sensitivity involved. The complexities of securing AI were a recurring theme, emphasizing the need for a holistic approach.
Collaboration and integration is the key to effective AI security
A compelling example of this collaborative approach was Cisco’s partnership with ServiceNow. This integration of Cisco AI Defense with ServiceNow SecOps promises improved threat response, but it also underscores the crucial need for close collaboration between product teams from different companies.
MCP and Zero Trust
The conference also explored other critical security topics. MCP (Model Context Protocol) and agent-to-agent communication present new challenges to security. One of the CISOs we spoke with during RSAC 2025 Conference called MCP the opposite of Zero Trust. We discuss how to interpret this statement and also how to deal with MCP from a security perspective.
It is difficult to talk about AI without also talking about APIs. Those deserve a lot of attention too, especially when it comes to security. API security should be part of bigger security platforms, but it is advisable to keep a close eye on its security.
LLMs to secure AI
One relatively recent trend in cybersecurity is the development of security-specific LLMs. Trend Micro recently open-sourced part of its Cybertron offering, and Cisco announced a similar LLM earlier this week too. Making these models available to the open-source community is potentially interesting. It could increase adoption and the overall development of those models.
Preparing for quantum
During RSAC 2025 Conference we had various discussions on what’s next, after current and future security challenges with AI. Quantum computing is something that may even have a bigger impact on security than AI. While there’s a general consensus on the need for post-quantum cryptography, the road towards getting there isn’t obvious for many companies. The best way to go about this is to do it in smaller segments. Don’t try to it all in one big project, that’s too much to take on.
Emerging trend: Browser Detection and Response (BDR)
At shows like RSAC 2025 Conference, we also always run into relatively new topics too. This year, we sat down with the founder of SquareX, a company that develops a Browser Detection and Response (BDR) solution. With workloads increasingly shifting to browsers and existing endpoint detection response solutions often proving insufficient, this could be an area of significant growth. However, the viability of browser-based detection and response as a standalone business remains uncertain.
More Techzine Perspectives:
The Techzine Perspective: Google Cloud puts all its eggs in the AI basket
The Techzine Perspective: Atlassian Team ’25 highlights power of integration
Techzine Perspective: The unexpected return of the VM at KubeCon EU