RSAC 2025: Why does the Ai Agent era mean more CISOS search
Join our daily and weekly newsletters for the latest updates and exclusive content of a leading AI coverage industry. Learn more
While over 20 suppliers announced agents, applications and security platforms based on AI RSAC 2025The most productive news from the conference is a rare, promoting tendency for security leaders. For the first time in three years, the overall effectiveness of cybersecurity has improved.
Scale of Venture Partners (SVP) recently released Cybersecurity Perspective Report 2025, which shared that the average efficiency of cybersecurity protection was improved for the first time in three years, increasing to 61% efficiency this year from 48% in 2023. According to the report “70% of security leaders were most protected against common phishing attacks, with only 28% of the company reports.”
SVP also found that 77% of CISO believes that the protection of AI/ML models and data pipelines was a priority to improve their security posture by 2025, compared to 55% last year. In particular, given the influx of new agent AI solutions announced in RSAC, 75% of the companies expressed interest in using AI to automate SOC investigations using AI agents to trigger large volumes of security alerts to prevent security incidents.
Increasing the number of efficiency of SVP is not accidental; They are the result of CISOS and their teams receiving automation on a scale, while successfully consolidating their platforms and reducing the gaps that the attackers have passed in the past.
“If you do not have full visibility, the attackers will go through the cracks between the products,” said Etay Maor, senior director of the CATO Networks Security Strategy, “VentureBeat told RSAC 2025.
Agentic AI moves quickly beyond a minimal viable product to the DNA platform
MAOR’s prospect explains why a new definition of what a minimal viable product for agency AI in cybersecurity is needed. RSAC 2025 revealed how mature AI is. There are a group of suppliers who use agent AI as a codenic adhesive to combine code bases and applications together, and then there are those that have been on this for years, and the AI agency is a major for their code base and architecture.
Cybersecurity suppliers in this last group, where the AI agent is a major for their platform and in many cases continue to double their research and development activity for distinguished AGentic AI. This includes SASE’s cloud platform of Cato Networks., Protection of Cisco AI., Single Agency Architecture of Falcon on Crowdstrike., Cyber Ai of Darktrace’s Cyber Ai., Elastic’s elastic AI assistant., Microsoft Security Copilot and Defender XDR, Palo Alto Networks Xsiam Cortex., Sentinelone’s Singularity Platform and VECTRA AI Discovery PlatformS
Organizations that rely on an integrated AI-rifled detection with automated restriction reduce stay time by completing 40%S They are too almost twice as likely To neutralize phishing based invades before lateral motion occurs. Sellers on the show of the show often rely on identity and access scenarios to show how their AI agent work flows can help cut the workloads of the Security Center (SOC) loads.
“The identity will be a critical element of AI throughout its life cycle. A -agents will need identities. They will have to understand zero confidence and how to check them? They explicitly manage the lowest privileged access,” said Microsoft’s Corporate Vice President for Security. As Jackkal said, “AI must first start with security. It is crucial to develop our security mechanisms as fast as we develop AI.”
A common topic of any demonstration of Agent AI in the show show was a triangulation of attack data, a quick acquisition of an idea of the form of the commercial operator used, and then defining a real -time restriction strategy.
Crowdstrike has shown how the AI agent can target real -time detection by live investigation into a North Korean campaign to threaten DEVOPS hires in US and worldwide strategic technology companies. The Demo live followed the merchant ship of the DPRK’s famous chollima As he posed as a remote Devops rent, he slipped along the HR checks and uses legitimate tools, including RMM software and VS code to erupt quietly. It was a sharp reminder that, although powerful, the agent AI still relies on a person in the cycle to notice adaptive threats and models of fine tuning before the signal was lost in the noise.
GEN AI’s goal: opening national commercial operations and killing it
This is the attacks that no person, company or nation sees that they are coming, which are the most pagan and challenging to contain and overcome. The thought of threats so devastating that they could easily close the power line, payment, banking or chain supply system dominates the minds of many of the most innovative and innovative technologies in cybersecurity.
Cisco Jeetu Patel’s chief product director emphasizes the urgency of reinforcing cybersecurity with AI, so the threats they are lurking, which can be devastating once triggered can be found now and neutralized. “AI is changing everything dramatically, and cybersecurity is at the heart of this. We are no longer dealing with human scale; these attacks appear on a machine scale,” Patel said during his main note.
Patel stressed that models managed by AI have not been determined: “They will not give you the same answer every time, introducing unprecedented risks.”
Ciso should understand today’s complex risks and threats
“This is not another AI talk, I promise,” jokes Crowdstrike Executive Director George Kurtz when he opened his foundation to RSAC 2025. “I was asked to give one and said,” How do you say about something that actually, such as a CISOS place on the table, Rotation of the decisive question of the guide of cybersecurity in 2025.
In his main note, “Ciso’s management to secure headquarters on board” Kurtz issued a clear call to actIon: “Cybersecurity is no longer a proposal for compliance. This is a mandate for management. The boards do not develop only; They are forced to comply with cyber risk as a major business threat.
Kurtz supported his argument with solid numbers: 72% of the tips say they are actively looking for cybersecurity experience, but only 29% actually have itS “It’s not just a difference in talent,” Kurtz said. “This is an opportunity if you are ready to activate,” he encourages.
His CISO roadmap reaches the meeting room was tactical and practical:
- Align the power of your business. “Find out where the business value is created. If you can’t talk margin, ARR or legal risk, you will not last long on the table.”
- Speak the language on board. “Each meeting room works on three priorities: time, money and legal risk. If you cannot translate cyber into them, you will stay on the sidelines.”
- Build your brand outside the security balloon. “The members of the board are on many advice. The road is through trust and reputation, not just technical achievements.”
Kurtz traced the way from the regulatory reform to the impact of the meeting room, reviewing how Sarbanes-Oxley in 2002 turned financial directors into solid associates in the meeting room. He claims that the term of office for reporting SEC violations in 2024 did the same for CISO. “The threats stimulate the regulation and regulation drives the composition of the board,” he said. “That’s our moment.”
His advice was not abstract. He urged Ciso to study proxy statements, identify the needs of the committee level, and to contact strategically with the board members who “always seek to play roles.” He pointed to Crowdstrike Ciso Adam Zoller, now on the Aventhealth board as a model. Zoller, Kurtz says, is someone who won his place by staying in the room, learning how the board has operated and is considered more than a security expert.
Kurtz closed with a challenge: “I hope to come back in ten years, still with red hair, and see Cisos at 50% of the boards, just like financial contractors. The regional hall is not waiting for permission. The only question is: Will you be?”
“AI is not magic – mathematics”
Diana Kelly, CTO of To protect youDraw one of the most significant early crowds in RSAC 2025 with a dumb message: “AI is not magic – mathematics. And just as we provide software, we must strictly provide AI’s life cycle.” Its main note provided a sound background that cut through the Gen Ai Hype, projecting the real risks for AI models that each organization must protect itself before any work on its models begins. Kelly provided an in -depth idea of patterns, quick injections and hallucinations, calling for a full -stack approach to AI security.
She introduced the Owasp Top 10 for GEN AI, emphasizing the need to provide AI by Day Zero, a partner with CISOS early, aggressively threatening and treatment of promotions, outputs and chains of agents such as privileged attacks of attack.
Declared Palo Alto Networks His intention to acquire Protect AI On the same day as Kelly’s presentation, another driving factor So many conversations about her main note.
RSAC 2025 indicates why it’s time to get the agent AI to give results
The RSAC 2025 made one thing clear: AI agents are entering the work processes of security, but the boards want proof that they are working. For CISO, under pressure to justify costs and reduce the risk, the focus is transferred from over -innovation to surgical impact. True wins, including 40% more habitat and phishing resistance reaching 70%, come from the consolidation of the platform and automation of a warning triage, which are all proven technologies and techniques. The moment of the agent’s truth is here, especially for suppliers who are just entering the market.
English