New Content

AI Red Teaming - Initial Access Collection

Release date: May, 6th, 2026

We’ve found the models; now it’s time to establish a foothold. This week’s release focuses on the Initial Access tactic, moving from observation to active exploitation within AI-integrated environments.

This new collection demonstrates how adversaries gain their first foothold in a system by targeting the way AI models ingest and process data. We aren't just talking about simple text prompts—this is about Initial Access via the poisoning of data sources (PDFs, Web crawlers, or Emails) that a RAG pipeline or AI agent will later execute.

MITRE ATLAS Techniques Covered:

AI Supply Chain Compromise (AML.T0010): Establishing access by poisoning "upstream" data sources before they reach the model.
Prompt Infiltration via Public-Facing Application (AML.T0093): Using public interfaces to bypass safety guardrails and gain an initial foothold.

Why

This is the #1 concern for enterprise customers today. As RAG pipelines and AI Agents are deployed with direct access to sensitive data and internal APIs, the risk of "Excessive Agency"—where the AI performs unauthorized actions on behalf of the attacker—is a critical vulnerability. These labs prove exactly how that initial breach occurs, providing a high-value differentiator for security teams who need to defend these new perimeters.

Who is this content for?

Experienced Adversarial Researchers ready to prove tangible risk in AI environments.
Red Teamers focused on gaining footholds via autonomous agents.
Security Architects designing the security boundaries for RAG and LLM integrations.

This content is available to all customers with the Immersive Labs AI Labs package.

AI Blue Teaming - Introduction

Release date: May, 6th, 2026

As Artificial Intelligence (AI) becomes deeply integrated into the enterprise, it introduces a new class of security threats. This lab introduces the fundamental skills required to detect, investigate, and contain incidents involving AI. This lab introduces the fundamental skills required to detect, investigate, and contain incidents involving Large Language Models (LLMs) and AI-driven applications. It looks at frameworks such as MITRE ATLAS and NVIDIA Kill chain, as well as the Regulatory Requirements Driving AI Security.

Why

The rapid adoption of generative AI has outpaced traditional incident response frameworks. We are releasing this collection now to help your organization:

Address the "Shadow AI" risk: As employees and departments deploy AI tools independently, your team needs the skills to respond to incidents in environments that sit outside of legacy security controls.
Mitigate high-impact threats: AI incidents often involve sensitive data exposure or automated logic manipulation; mastering these fundamentals is critical to preventing significant reputational damage.
Future-proof your SOC: By establishing an AIIR capability today, you ensure your team is prepared for the next generation of automated, AI-driven adversary tactics.

This is the first of a number of Blue team collections that will align with our recent Red team AI content, to enable organizations to transform their capabilities across their security function and move towards AI purple team.

Who is this content for?

Security Operations Center (SOC) Analysts: To learn how to triage alerts that originate from AI enabled solutions.
Incident Responders: For those who need a dedicated playbook for containing AI-driven attacks without disrupting critical business automation.
Application Security (AppSec) Engineers: To understand how vulnerabilities in AI integration points translate into active security incidents.

This content is available to all customers with the Immersive Labs AI Labs package.

Links to the lab: part 1 and part 2.

New GRC Collections

Release date: May, 6th, 2026

We're releasing two new collections as part of our Governance, Risk, and Compliance content.

SOX Cyber and IT Controls Financial integrity and IT security are now inseparable. This collection connects the dots between the Sarbanes-Oxley Act and the IT General Controls (ITGCs) required to prevent modern financial fraud.

Target: Internal Auditors, Finance Managers, and IT Leads.
The "Why": Modern financial fraud increasingly relies on IT vulnerabilities, making "IT General Controls" a primary focus for auditors.

SOC Reporting Everything you need to know about SOC 1, 2, and 3. This collection explains how these reports build market trust and why they are vital for the modern sales cycle.

Target: Sales Professionals, Vendor Risk Managers, and BIZDEV.
The "Why": SOC 2 is now the "de facto" requirement for any B2B SaaS partnership.

Identity for AI agents Collection

Release date: May, 13th, 2026

Establish a verifiable and secure identity framework for autonomous agents to mitigate the risk of unauthorized access and spoofing within machine-to-machine ecosystems.

What are we doing?

We are releasing a new collection titled Agent Identity to help you master the security of autonomous interactions. This collection focuses on the Agent-to-Agent (A2A) protocol, which is the foundational framework that allows artificial intelligence (AI) agents to identify, authenticate, and communicate with one another securely. You will learn how to implement robust identity standards that prevent unauthorized agents from accessing your sensitive data or spoofing legitimate services.

Who are we doing it for?

This collection is designed for security engineers, AI developers, and system architects who need to manage the risks associated with automated ecosystems. If you're responsible for deploying autonomous agents or securing Machine-to-Machine (M2M) communications, these labs provide the technical depth required to ensure your agents are trustworthy and accountable.

Why are we doing it now?

As organizations increasingly rely on AI to automate complex workflows, the surface area for identity-based attacks has expanded. Standard human-centric Identity and Access Management (IAM) is no longer sufficient for the speed and scale of agent interactions. By launching this now, we're providing you with up-to-date information on how to treat AI agents as first-class citizens in your security architecture, ensuring every automated request is verified and logged.

Business Continuity Collection

Release date: May, 13th, 2026

What are we doing?

We are launching a new collection called Business Continuity: Recovery & Returning to Normal Operations: This seven labs collection moves beyond the initial crisis response to focus on the complexities of returning to normal operations. You will explore what “recovery” actually means in a business context, establish robust recovery governance and assurance, and master communications during the recovery phase. Practical labs also guide you through managing backlogs, bottlenecks, and operational debt, as well as planning the delicate transition out of continuity mode while supporting the human side of recovery.

Who are we doing it for?

This collection is designed for senior leaders, department heads, and BCM professionals who are responsible for the long-term stabilization of an organization after a disruption. It is also essential for HR partners and communications teams who need to manage the well-being of staff and the reputation of the business during the transition back to "business as usual."

Why are we doing it now?

Many organizations fail not during the crisis itself, but during the messy and often overlooked recovery phase. We are providing this content now to ensure you have the technical exercising capabilities to manage the full lifecycle of a disruption. By mastering these recovery principles today, you can eliminate operational debt faster and ensure a 100% successful return to standard operations without burning out your workforce.

New GRC Lab Releases: Global Resilience and Modern Risk

Release date: May, 20th, 2026

Today we’re dropping the next wave of our GRC catalogue refresh. As we move through 2026, the regulatory landscape is shifting from static checklists to active, enforcement-driven operational resilience. This release bridges the gap between core risk theory and massive new global compliance frameworks.

NIS2 Directive

What's new: The EU's Network and Information Security Directive has officially expanded its reach. This collection breaks down the stricter enforcement regimes and clarifies the critical distinction between "Essential" and "Important" entities.
Target Audience: Legal Teams, Supply Chain Managers, and EU-Based Leaders.
Why: As EU member states finalize local laws this year, businesses must immediately identify if they fall under these new compliance mandates to avoid massive penalties.

DORA (Digital Operational Resilience Act)

What's new: A guide to the EU framework that forces financial institutions to prove they can withstand, respond to, and recover from severe ICT disruptions.
Target Audience: Financial Services Leadership, Risk Officers, and Third-Party ICT Providers.
Why: Enforcement is officially live. The regulatory focus has permanently shifted from traditional "financial" buffer zones to true "digital" operational resilience.

CCPA Foundations

What's new: A dive into the California Consumer Privacy Act, mapping out the precise rights of California consumers and the strict operational obligations placed on businesses handling their data.
Target Audience: Marketing Teams, Data Analysts, and US-Based Customer Service Leads.
Why: Continuous amendments and aggressive enforcement by the CPPA have significantly heightened the risk of non-compliance for modern businesses.

New Dynamic Threat Range: Nebula Banking - Event Horizon

Release date: May, 20th, 2026

Today we've published a new Dynamic Threat Range centred around a multi-stage intrusion against a financial services provider. This scenario covers AI chatbot exploitation via prompt injection, container escape through Docker socket abuse, Active Directory certificate abuse (ESC1), and financial fraud, giving defenders hands-on experience tracing a realistic attack chain from initial access to objective completion.

Who is this for?

This range is designed for SOC analysts, incident responders, and threat hunters who need experience investigating modern hybrid attack chains. It's particularly relevant for teams defending environments that combine AI-powered applications, containerised infrastructure, and traditional Active Directory — the kind of architecture increasingly common in production but underrepresented in training content.

Why are we doing it now?

As organisations adopt GenAI tools in customer-facing applications, the attack surface has expanded beyond traditional network exploitation. Attackers are leveraging prompt injection, function abuse, and command injection against AI interfaces to gain initial footholds, then pivoting through container infrastructure into corporate networks. This range provides up-to-date training on detecting these emerging attack paths across Elastic, Splunk, and Microsoft Sentinel SIEMs.

New OT Collection - Incident Command

Release date: May, 27th, 2026

We are launching the OT Incident Command collection, which consists of seven labs designed to bridge the gap between technical detection and executive leadership in industrial environments. This collection introduces you to the leadership and coordination aspects of managing complex Operational Technology (OT) security incidents. The curriculum follows the Incident Command System for Industrial Control Systems (ICS4ICS) framework, ensuring you can manage stakeholders like plant managers and safety engineers while balancing security with production uptime. The labs conclude with a Workforce exercise, allowing you to synthesize these advanced capabilities in a realistic scenario.

Who is this for?

This collection is designed for "on-the-ground" decision makers and managers. These labs will give an operations manager or a security manager the practical skills you need to lead teams effectively.

Why are we doing it now?

When an OT incident hits, response operations can easily stall without a clear structure. Success depends on coordinated leadership on the ground, where every second counts. We're releasing this collection now to help organizations build structured leadership skills before an incident occurs, ensuring teams can respond rapidly, communicate clearly, and minimize downtime.

New Lab Collection: AI Red Teaming — Persistence

Release date: May, 27th, 2026

Gaining an initial foothold is only half the battle. This week, we are moving into the Persistence tactic of the MITRE ATLAS™ framework—showing how adversaries ensure they don't lose access to AI systems across restarts, changed credentials, or network interruptions.

This new collection consists focused on the long-term state of an AI environment. Instead of simple, session-isolated exploits, these labs demonstrate how an attacker can manipulate an AI agent's memory or poison its integrated tools to maintain a stealthy, long-term presence.

MITRE ATLAS™ Techniques Covered:

AI Agent Tool Poisoning (AML.T0110): Weaponizing the tools and descriptions an AI agent relies on to interact with its environment, forcing malicious execution.
AI Agent Context Poisoning: Memory (AML.T0080.000): Injecting malicious instructions directly into an LLM's cross-session database, ensuring changes persist into future, unrelated user sessions.

Why is this critical now?

AI models are no longer standalone side-projects; they are fully integrated into core production CI/CD pipelines. If an attacker can achieve persistence within a deployment workflow (like a compromised code repository or an S3 bucket) or corrupt its cross-session memory, they can subvert every single iteration of the model that follows.

Who is this content for?

Adversarial Researchers testing the long-term resilience of production AI environments.
Red Teamers looking to prove that a breach in the pipeline allows an attacker to remain undetected for long periods.

Link to the AI Red Teaming collection can be found here.

New Lab Collection: AI Blue Teaming — Reconnaissance

Release date: May, 27th, 2026

Building on the introduction to our AI Blue Teaming framework , we are launching our first technical defensive collection. This release mirrors our previous AI Red Teaming: Reconnaissance Collection, shifting the lens to a blue-team perspective.
Before an adversary can exploit your AI infrastructure, they have to map it. This collection is about making sure they get caught in the act.

This collection is designed to teach defenders how to spot and intercept attempts to map an organization's AI attack surface. When adversaries try to sniff out where your customer-facing chatbots, internal RAG pipelines, or backend models live, these labs show your team how to deploy preventive mitigations and high-fidelity identification techniques to shut them down early.

Why are we doing it now?

The explosive adoption of generative AI has left traditional incident response frameworks in the dust. We are shipping this collection now to help organizations:

Stop the Attack Lifecycle Early: Catching an adversary during reconnaissance prevents them from tailoring model-specific exploits down the line.
Future-Proof the SOC: By establishing an AIIR (AI Incident Response) capability today, you ensure your security operations team is ready for the next generation of automated, AI-driven adversary tactics.

Who is this content for?

This collection is built for modern defence teams evolving to meet AI-driven threats:

SOC Analysts & Incident Responders needing to identify AI-specific reconnaissance telemetry.
Application Security (AppSec) Engineers implementing preventive mitigations around AI endpoints.
Purple Team Analysts looking to validate detection engineering use cases against AI recon techniques.

Link to the AI Blue Teaming collection can be found here.