Connect with us

Hi, what are you looking for?

SecurityWeekSecurityWeek

Incident Response

Is The SOC Obsolete, And We Just Haven’t Admitted It Yet?

Many AI-first enterprises have already embraced sovereign architectures for general AI initiatives; cybersecurity—and the SOC—should be next.

Security Operations Center (SOC)

For decades, the Security Operations Center (SOC) has been the beating heart of enterprise defense. Analysts monitor dashboards, triage alerts, and investigate incidents around the clock. The SOC is often portrayed as the last line of defense—a place where intelligence meets action. And yet, if we are honest, the SOC as we know it is already obsolete. Not because analysts aren’t skilled or diligent, but because the very nature of cyber threats has changed faster than our operational models can keep up.

The modern SOC is still largely a human-centric workflow. Analysts pivot between tools, manually enrich alerts, and painstakingly validate detection rules. Security vendors promise a revolution: an AI SOC capable of autonomous investigations, dramatically reduced workloads, and proactive threat response. In practice, these promises remain aspirational.

In fact, Anton Chuvakin and Oliver Rochford recently co-authored When Marketing Fails, highlighting the gap between AI SOC marketing claims and reality. Based on vendor interactions, practitioner interviews, and SOC community OSINT, their report shows that while AI can assist SOC analysts, it rarely replaces human effort or autonomously resolves incidents. Analysts remain the bottleneck, and AI often lacks the contextual understanding required to make fully reliable decisions.

Threat Actors Are Operating At Machine Speed

Meanwhile, attackers are no longer constrained by traditional human limitations. In late 2025, Google’s Threat Intelligence Group confirmed that cybercriminals are already deploying AI‑powered malware that rewrites and adapts its own code during execution, marking a watershed shift in offensive capabilities. One notable example, PROMPTFLUX, uses real‑time interaction with Google’s Gemini model to dynamically regenerate its VBScript payload to evade detection and persistence mechanisms mid‑attack — a level of autonomous adaptation unseen in conventional malware families.

Similarly, Anthropic reported disrupting what it described as one of the first large‑scale AI‑orchestrated cyber espionage campaigns, where an AI tool executed vast portions of an intrusion framework with minimal direct human intervention.

Data from Google/Mandiant’s M‑Trends 2026 report shows that attackers are accelerating their operational tempo across the board, with exploitation increasingly occurring before patches are published and lateral movement hand‑offs collapsing to mere seconds (22 seconds to be exact!). And of course, while Anthropic’s Mythos Preview is still in the hands of defenders, it has demonstrated the ability to surface hundreds of defects in days that would normally take elite researchers months, and chain multiple low-levels into a single critical exploit.

Advertisement. Scroll to continue reading.

This means adversaries are now operating at near‑machine speed — adapting, targeting, and executing campaigns faster than traditional human‑centric SOC workflows can keep up.

Can The Traditional SOC Keep Up?

The foundation of AI-driven defense is complete, unfiltered data. Security teams must be able to ingest and analyze every relevant signal—including sensitive information such as source code, internal documents, and privileged communications—without compromising privacy, security, or organizational sovereignty. In this context, sovereignty means that the cybersecurity stack, data, and AI models remain fully under the organization’s control, with no reliance on third-party multi-tenant platforms that limit access or impose policy constraints. Only by having unrestricted access to both historical and current datasets can AI be applied effectively, enabling accurate detection, deep correlation, and meaningful long-term trend analysis.

Once this foundation is in place, agentic AI capabilities—including explainability, auditability, and reproducibility—can function reliably. Analysts must understand why AI reaches specific conclusions, and every decision and action must be logged and reproducible for compliance and operational trust. Without full data access, these capabilities are superficial at best: AI decisions become opaque, incomplete, and prone to blind spots, leaving organizations vulnerable to fast, adaptive attacks.

The limiting factor is not human skill—it is architecture. Many SOCs still rely on cloud-based SIEMs or XDR platforms where storage and compute costs force analysts to filter, truncate, or delete data. Privacy and sovereignty concerns often prevent certain datasets from being sent to the cloud for analysis. This creates blind spots that attackers readily exploit.

Many AI-first enterprises have already embraced sovereign architectures for general AI initiatives; cybersecurity—and the SOC—should be next. When AI can operate on the full dataset, and when actions are explainable, auditable, and reproducible, analysts can finally move beyond human-bottlenecked workflows. The SOC transforms into a truly adaptive, machine-speed defensive engine, capable of matching and outpacing AI-driven adversaries.

The SOC Of the Future: Humans and AI Operating On Complete Data

The SOC of the future will invert the current model. Signals will feed AI agents continuously, correlations will be drawn automatically, and human analysts will focus on oversight, exception handling, and strategic response. Humans will no longer chase alerts—they will guide autonomous systems, investigate the hardest problems, and make high-stakes decisions with confidence.

The SOC is not failing because it is inherently flawed; it is obsolete because threats have evolved faster than the SOC architecture defending against them. Enterprises that acknowledge this reality and deploy data complete, AI-native security solutions for their SOC will be positioned to survive—and even thrive—against machine-speed adversaries.

Related: Cyber Insights 2026: Threat Hunting in an Age of Automation and AI

Written By

Danelle Au is a cybersecurity and AI go-to-market leader with 20+ years of experience bringing disruptive security, cloud, and AI technologies to market. She is currently VP of Product Marketing at Cylake. Danelle has held multiple CMO and VP roles across startups and market leaders—including Infoblox, Ordr, Blue Hexagon, SafeBreach, and Adallom—helping define emerging security categories and scale go-to-market engines. She is a co-founder and co-author, has multiple U.S. patents, and holds an M.S. in Electrical Engineering from UC Berkeley. The opinions and views expressed within her articles are those of Danelle alone in her personal capacity and do not necessarily reflect the positions of Cylake or any of her prior employers.

Trending

Daily Briefing Newsletter

Subscribe to the SecurityWeek Email Briefing to stay informed on the latest threats, trends, and technology, along with insightful columns from industry experts.

In cyber-physical systems (CPS), just one hour of downtime can outweigh an entire annual security budget. Learn how to master the Return on Security Investment (ROSI) to align security goals with the bottom-line priorities.

Register

Delve into big-picture strategies to reduce attack surfaces, improve patch management, conduct post-incident forensics, and tools and tricks needed in a modern organization.

Register

People on the Move

Malwarebytes has named Chung Ip as Chief Financial Officer.

Semperis has appointed John Podboy as Chief Information Security Officer.

Randy Menon has become Chief Product and Marketing Officer at One Identity.

More People On The Move

Expert Insights

Daily Briefing Newsletter

Subscribe to the SecurityWeek Email Briefing to stay informed on the latest cybersecurity news, threats, and expert insights. Unsubscribe at any time.