AI at RSAC: The innovations that will shape cybersecurity’s future

Article insights:

Key AI Innovations at RSAC – The most promising advancements in AI-driven security, from autonomous threat detection to adversarial AI defenses.

How to Separate Hype from Reality – How to critically evaluate AI claims from vendors and researchers to identify solutions with real-world impact.

The Future of AI in Cybersecurity – How emerging AI technologies will influence security strategies, regulatory frameworks, and adversarial tactics in the coming year.

Artificial intelligence is set to dominate the conversation at this year’s RSA Conference, but with so much noise, how can security leaders separate true innovation from hype?

In a pre-RSAC Hot Topic webinar, Security Weekly hosts Paul Asadoorian and Adrian Sanabria sat down with Sounil Yu, CTO and Co-Founder at Knostic, and explored the AI advancements that are set to make the biggest impact.

From cutting-edge threat detection and autonomous response to AI-driven security automation and adversarial machine learning, the session unpacked what to expect from vendors, researchers, and policymakers at RSAC.

The questions they sought to address:

Will we see real breakthroughs, or just more AI-washing?

How are enterprises adopting AI-powered defenses?

What risks should security teams prepare for?

Innovations ahead

The discussion highlighted several critical AI innovations the panelists believe are poised to transform cybersecurity:

Agentic AI Systems: Yu predicts 2025 will be the "year of agents" -- AI systems that can autonomously perform complex tasks across multiple platforms. These agents will move beyond simple language models to actively coordinate actions, using protocols like Multi-Agent Communication Protocol (MCP) to interact seamlessly across different tools and environments.

Contextual Reasoning and Need-to-Know Policies: A breakthrough approach involves developing AI systems with sophisticated contextual understanding. Instead of blanket access or denial, these systems will dynamically assess an individual's "need to know" - providing precisely tailored information based on job function, clearance level, and specific requirements.

Advanced Threat Detection and Inference: AI's capability to make nuanced inferences goes beyond traditional threat detection. These systems can now analyze subtle data patterns to potentially predict organizational changes, security risks, or strategic shifts before they become apparent through conventional monitoring.

Intelligent Access Management: Future AI systems will transform access control from a restrictive "no" approach to a more dynamic, permission-based model. They'll not just block access but suggest alternative information or pathways that meet security requirements while enabling business productivity.

Vulnerability Management Evolution: While current AI struggles with comprehensive vulnerability discovery, it shows promise in automating mundane bug tracking and remediation tasks. This could free security researchers to focus on more complex, creative problem-solving.

Challenges and considerations

The panel emphasized critical considerations for AI integration:

Maintaining human oversight

Establishing robust governance frameworks

Developing granular access controls

Preventing potential misuse or over-privileged AI agents

The panel stressed that AI should augment human capabilities, not replace them entirely -- helping professionals "wash dishes" so they can focus on more strategic, creative work.

As RSAC approaches, these AI innovations represent a significant leap in cybersecurity's technological capabilities, promising more intelligent, adaptive, and context-aware security solutions.