The honest view from inside technical leadership
AI News

AI News

Daily AI news with an engineering leadership take, each story credited to its original AI News report.

Security · AI News

EY’s Agentic Cyber Defense Center Records the Authority Behind Every Security Agent Action

EY is selling agentic security operations with a governance layer that writes down what each agent may do and keeps a record of what it did. The record is only as useful as the rules your own teams write for it.

Line-art illustration of a shield linked to a chain of sealed record blocks and a checklist, representing the EY Agentic Cyber Defense Center governing AI security agents

EY launched the EY Agentic Cyber Defense Center on November 18, 2026, a security operations platform built on Microsoft technology that puts policy-based controls and tamper-evident records around the actions of AI security agents. EY says the platform is meant to show how each agent acted, under whose authority, and who is accountable, while people keep the final say on consequential decisions.

The EY Agentic Cyber Defense Center Adds a Governance Layer for Agent Authority

The core of the EY announcement is a governance layer that EY says defines what each agent is authorised to do and records what it actually does. EY packages this with its EY.ai Value Blueprints, which it describes as a structure linking cyber-domain agents, reusable integration patterns, guardrails, human decision rights and what it calls an agentic evidence chain. The blueprints span identity, data, applications, cloud and operational technology. Richard Watson, EY’s global consulting cybersecurity leader, said security now needs “an entirely new operating model.” EY published no pricing, customer names or performance figures with the launch.

Microsoft Supplies the Agents, and EY Says Their Availability Varies

EY lists three Microsoft capabilities inside the platform. Codename MDASH uses several AI models together to scan code for vulnerabilities. Microsoft Perception coordinates red, blue and green team agents for attack simulation, detection, triage and remediation. ISOC in Microsoft Defender combines investigation, response and protection. EY’s release says these are subject to product availability. Microsoft’s own November 17 Ignite post lists Perception and core ISOC capabilities as generally available, while MDASH remains in an expanding preview that customers must be nominated for. EY adds that deployments can run in the client’s own cloud region or fully on premises, which matters for organisations under data residency or sovereign AI rules.

EY Tested the Platform on Itself as Client Zero

EY says it is one of the first organisations to roll out Microsoft security for agentic cyber defense across its own global operations, and it uses that rollout to validate operating models and codify controls before selling them. Writing for AI News, Ryan Daws makes a point worth holding onto: a tamper-evident log lets a team reconstruct what an agent did after the fact, and whether the agent chose well is a separate question. The quality of the outcome still depends on the rules that decided what the agent was allowed to do in the first place.

Why This Matters for Engineering Leadership

The useful idea in this launch is that agent authority becomes a written, enforced policy with an audit trail, and engineering teams own most of the systems those policies touch. A security agent that revokes access, rolls back a cloud change or blocks a deployment is acting on infrastructure your platform and application teams run. Before any agentic SOC engagement starts, whether with EY or another provider, write the decision-rights table yourself: which actions an agent may take alone, which need an on-call engineer’s approval, and which are off limits, such as touching production databases or protected branches. Ask the vendor to show you a real record of an agent action, where those records are stored, who can read them, and what makes them tamper-evident. Check which named capabilities are generally available today, since a preview tool like MDASH should not gate releases. And put a platform engineer in the room when containment and remediation playbooks are written, because the first false positive will land on their pager.

FAQ

What is the EY Agentic Cyber Defense Center?

It is an EY security operations platform built on Microsoft security technology, announced on November 18, 2026. EY says it applies policy-based controls and tamper-evident records to the actions of AI security agents.

Which Microsoft tools does the platform use?

EY names Codename MDASH for multi-model code vulnerability scanning, Microsoft Perception for coordinated red, blue and green team agents, and ISOC in Microsoft Defender. EY says these are subject to product availability.

Can the platform run outside a public cloud?

Yes. EY says deployments can run in the client’s own cloud region or fully on premises to meet sovereign AI requirements.

What does Client Zero mean in this announcement?

EY says it has deployed Microsoft security for agentic cyber defense across its own global organization first. It is using that rollout to validate operating models and codify reusable controls before client work.

Do tamper-evident records prove an AI agent acted correctly?

No. They show what an agent did and under which authority, so people can review it afterwards. Whether the action was right depends on the policies and approval rules the organization set.

Originally reported by Ryan Daws for AI News, November 19, 2026: EY and Microsoft launch agentic cyber defence centre. Primary source: EY newsroom, “EY announces EY Agentic Cyber Defense Center to help AI-powered enterprises scale secure operations” (November 18, 2026); Microsoft Security Blog, “Security for the agentic era, powered by Microsoft Perception” (November 17, 2026).

Read more AI News