AI agents are rapidly becoming embedded in everyday business operations, helping employees automate workflows, access information, and accelerate decision-making. As organizations embrace agentic AI, security teams face a growing challenge: understanding how AI is being used, what agents can access, and where risk may emerge.

As agents take on more business-critical work, security teams often need to move across multiple portals to understand risks spanning identity, data, and threat activity. This fragmented view can make it difficult to assess an agent's overall risk posture and determine where attention is needed. Organizations need a way to bring these signals together without disrupting existing security investments or workflows.

Today, Darktrace is announcing an integration between Darktrace / SECURE AI and Microsoft Agent 365 that brings Darktrace's Adaptive AI-driven risk signals directly into the Microsoft 365 Admin Center. By extending the visibility and risk understanding provided by Darktrace / SECURE AI into the Microsoft ecosystem, organizations can gain a more unified understanding of AI agent risk across their environments.

As one of the first security companies to partner with Microsoft to contribute third-party risk signals to the Agent Registry, Darktrace is helping shape how organizations understand and manage AI agent risk.

Extending visibility into the Microsoft Agent 365 experience

Microsoft Agent 365 provides administrators with a centralized registry of AI agents operating within their environment. As organizations expand their use of AI agents, this centralized visibility becomes increasingly important for governance and oversight.

This new integration extends that visibility by allowing Darktrace-generated risk signals to be surfaced directly within the Microsoft Agent 365 experience. By combining Microsoft's agent management and security capabilities with Darktrace's AI-powered risk analysis, organizations gain greater awareness of potential security concerns associated with AI agents operating across their environments.

By integrating Darktrace telemetry into Agent 365, customers can:  

• Surface Darktrace-detected risks and signals alongside Microsoft-native signals in a single interface

• Identify potentially compromised or anomalous AI agents more quickly

• Gain unified understanding of context and agent behavior  

This approach reinforces a single control plane for AI security while allowing organizations to continue leveraging existing investments across both platforms.  

Why unified visibility of AI agent risk signals matters

As AI adoption accelerates across Microsoft environments, organizations must manage new forms of behavior, access patterns, and risk. Security teams need more than inventories and permissions. They need visibility into how AI systems operate and how risk evolves over time.  

This integration addresses a critical gap: how to bring behavioral AI security insights into the same control plane as identity, access, and agent management.  

With Darktrace and Microsoft Agent 365 together, organizations benefit from:  

• Unified visibility: A single pane of glass for understanding AI agent risk signals across Microsoft and Darktrace signals  

• Faster detection of abnormal agent behavior: Darktrace's Adaptive AI highlights deviations that may not be captured by static controls  

• Operational efficiency: Security teams can triage and prioritize risk signals without switching between systems

• Stronger trust in AI deployments: Clear attribution, context, and investigation pathways improve confidence in AI agent usage

Extending Microsoft's AI security model, not replacing it

Securing AI requires a layered approach that combines governance, visibility, threat detection, and risk management. This integration is designed to complement Microsoft's security capabilities, not duplicate them.  

Through Darktrace / SECURE AI, Darktrace contributes:  

• Identification of risk via advanced prompt analysis  

• Behavioral anomaly detection across AI agents  

• Cross-environment threat correlation  

• Autonomous insight into emerging or unknown risks  

Microsoft provides:  

• Centralized agent management

• Identity and access governance

• Native detection of risk signals and enforcement capabilities

Together, these capabilities create a more complete, layered approach to securing AI-driven enterprises. Organizations gain the governance and policy controls needed to manage AI adoption while benefiting from continuous visibility into how AI is used across the business.

Building the future of secure AI

As AI agents become more deeply embedded in business processes, organizations need more than inventories and static controls. They need to understand how AI is being used, how agents behave, and where risk is emerging across the enterprise.

Darktrace / SECURE AI delivers that understanding through continuous visibility into AI activity, helping security teams assess intent, identify behavioral drift, and uncover emerging risk across both human and agent-driven workflows. Powered by Adaptive AI, it provides the context needed to secure AI as it evolves.  

The integration with Microsoft Agent 365 extends those insights into the workflows organizations already use. Agent 365 provides a unified control plane for governing and securing AI agents, while Darktrace contributes complementary behavioral risk signals that can be surfaced within the Agent 365 experience. Together, they give customers broader context on agent activity and risk while preserving the value of their existing Microsoft and Darktrace investments.

As enterprises move from AI experimentation to AI-powered execution, Microsoft and Darktrace help bring together governance, compliance, behavioral understanding, and oversight in a unified approach to AI security. For organizations adopting Microsoft 365 E7, Darktrace / SECURE AI further strengthens that foundation by providing continuous visibility into AI activity, agent behavior, and emerging risk as AI adoption scales.

In early 2026, a React2Shell honeypot purpose-built by Darktrace analysts was compromised in less than two hours after deployment. That single data point captures the pace of the threat landscape in the first half of 2026, but speed tells only part of the story.

The shift over the past six months has moved away from traditional malware and vulnerability-centric attacks and toward the abuse of trusted identities, platforms, and infrastructure. Identities, Software-as-a-Service (SaaS) platforms, cloud entitlements, automation frameworks, and non-human identities have become the preferred attack paths as organizations adopt AI at scale.

Attackers are increasingly operating inside the relationships, services, and authenticated channels that defenders and users are conditioned to rely on, rather than breaking in from the outside.

What has changed since 2025?

In 2025, identity became the new perimeter as attackers increasingly bypassed traditional exploitation in favor of trusted accounts, SaaS platforms, and emerging AI-enabled tradecraft. The first half of 2026 marks the next stage of that evolution. Identity remains central, but the trust challenge now extends far beyond accounts to email authentication, cloud entitlements, software supply chains, AI gateways, remote administration tooling, and non-human identities.

Identity and email: trust signals under pressure

Email remains the most reliable route to a trusted identity, and the data shows attackers investing in quality over noise. In the first half of 2026, 67% of phishing emails passed DMARC. Authentication alone is no longer sufficient to stop most phishing attempts. VIP users were targeted in 25.8% of phishing, consistent with 2025's “over 25%” figure, but drifting upward throughout the period. Crucially, phishing sophistication continued to increase: 37% of phishing contained a high volume of text, up from 32% in the first half of 2025, while 39% featured novel social engineering techniques, suggesting attackers are further customizing to specific targets.

The most prevalent threats affecting Darktrace customers were also among the most identity-centric: information stealers, with dedicated StealC and AMOS campaigns running through the half-year. Their prevalence is, fundamentally, an identity story. Credentials harvested by infostealers often become the initial access vector for far higher-impact intrusions later in the attack chain. Crucially, the delivery method rarely requires exploitation of a technical weakness. ClickFix social engineering, which tricks users into running malicious code themselves, remained a common distribution route. One recent campaign impacted Darktrace customers across 17 countries, with the United States the most affected. The compromise did not begin with a software flaw, but with a trusted user taking a trusted action.

Supply chain: Trust weaponized at scale

March and April reinforced the same lesson: trust has become a supply-chain vulnerability. The Axios compromise abused trust in a widely used maintainer, while the Trivy campaign leveraged trusted CI/CD infrastructure, release artifacts, and container images to push malicious code through legitimate development workflows.

The clearest example was a February–March campaign in which devices downloaded malicious payloads while using Hola VPN, later linked to an issue within Hola's own delivery pipeline. Darktrace's Threat Research team identified associated activity through recurring anomalous behavior across multiple customers before a public advisory was released.

More recently, attackers abused legitimate blockchain infrastructure to distribute infostealers, including AMOS and Phexia. Popular tools like VPNs, often used by users with limited security resources, combined with legitimate command-and-control (C2) infrastructure enables attackers to reach a far wider victim base while frustrating defenders who cannot simply block the associated endpoints.

For defenders, the challenge is no longer identifying malicious infrastructure, but recognizing when trusted infrastructure begins behaving maliciously.

Cloud and SaaS: from target to terrain

Through May and June, activity involving device registration, cloud data theft, SaaS abuse, RDP expansion, and remote management tooling suggested that attackers increasingly view cloud and SaaS not simply as targets, but as their preferred operating environment.

In one Darktrace case a single compromised SaaS account triggered activity across email, SaaS, and network layers, including inbox rule changes, phishing propagation, and connections to suspicious infrastructure. None of these indicators were decisive in isolation, but together they revealed a clear intrusion. Increasingly, attackers do not need to bypass trust controls in these environments; they inherit them through compromised identities, delegated access, and legitimate administration tools. This is the natural progression of 2025's SaaS-targeted ransomware trend: the platforms on which businesses operate are increasingly the same platforms on which adversaries operate.

AI: accelerant, attack surface, and trusted but risky actor

If trust is the attack surface, AI is where that surface is expanding fastest. Across the Darktrace customer base, AI service connections per deployment rose 13% in the first half of 2026, surpassing 16 million connections, while the typical organization now interacts with seven different AI providers. AI is no longer at the edge of the enterprise; it is embedded in day-to-day business operations. That shift creates three distinct problems, all of which were observed by Darktrace in the first half of 2026.

1. AI as an attack multiplier

Darktrace identified AI-generated malware exploiting React2Shell, in which an attacker used an LLM to produce working exploit code and deploy it at scale. Similar activity is increasingly appearing across the wider threat landscape, suggesting that the barrier to effective offensive operations is collapsing. As demonstrated by the recent JadePuffer case, in which an agentic threat actor exploited a vulnerability in an internet-facing server before launching a fully automated ransomware attack, AI is accelerating the path from vulnerability disclosure to operational exploitation [1].

2. AI as an attack surface

The AI layer itself is now worth probing. At an automation technology manufacturer, a compromised LLM proxy was used as a steppingstone toward additional AI services; when that failed, the attacker pivoted to cryptomining. Darktrace’s Cyber AI Analyst pieced the intrusion together and Darktrace’s Managed Threat Detection service alerted the customer, containing it before it could progress further. The practitioner lesson is clear: treat AI gateways, proxies, and model endpoints as production cloud workloads because attackers already do.

3. AI as a trusted but potentially risky actor

Darktrace / SECURE AI observations suggest the most common real-world risk is quieter still: employees entering personal identifiable information (PII), tax records, identity documents, company financial data, HR records, and personal medical data into LLM prompts, alongside widespread shadow AI use and increased AI usage from mobile devices. Across nearly 280,000 prompts submitted by almost 28,000 users over 28 days, Darktrace identified that approximately 1% of these prompts (or 2,945 instances) contained sensitive data*.

_{*Prompt data was analyzed in aggregate and anonymized form to protect user privacy.}

‍

For defenders, the challenge is context: knowing when legitimate business use crosses into material risk without breaking privacy or user trust. As organizations increasingly trust AI systems to access, process, and share sensitive information at machine speed, AI must be secured and monitored alongside identities, applications, and cloud infrastructure.

Speed and geopolitics: faster operations, longer-term objectives

Several investigations in the first half of the year showed how quickly attackers operationalize newly disclosed vulnerabilities, validating exploitation through Out-of-Band Application Security Testing (OAST) infrastructure and trusted cloud services before patching cycles can be completed. React2Shell was compromised in two hours, while BeyondTrust exploitation followed in less than a day. Against this backdrop, state-aligned actors continue to prioritize long-term access, intelligence collection, and pre-positioning through legitimate services, cloud infrastructure, and trusted relationships. Operations linked to China, Russia, Iran, and the Democratic People’s Republic of Korea (DPRK) shared a common characteristic: a focus on persistence and strategic positioning rather than immediate disruption.

China: Darktrace observed Chinese-nexus actors prioritizing long-term access through trusted services, dynamic-link library (DLL) sideloading, and modular intrusion chains consistent with activity documented in Crimson Echo reporting and associated with Twill Typhoon tradecraft.

Iran: Darktrace's ZionSiphon investigation highlighted Iranian-linked interest in operational technology (OT) environments, blending espionage objectives with infrastructure disruption capabilities.

Russia: Darktrace investigations, alongside wider industry reporting, highlighted Russian reliance on trusted relationships and supply-chain targeting for long-term intelligence on Ukraine related support [2].

DPRK: Darktrace observed DPRK-linked activity combining rapid vulnerability weaponization with persistent access techniques, including Axios supply-chain compromise, React2Shell exploitation and stealthy macOS intrusions

While objectives differed across actors, the tradecraft was remarkably consistent: trusted services, legitimate infrastructure, and persistent access remained more valuable than immediate disruption.

The defender shift

Across identity compromise, supply-chain attacks, SaaS abuse, AI infrastructure targeting, and state-aligned operations, attackers increasingly succeed by operating through trusted systems rather than breaking through defensive controls. Trusted users, trusted software, trusted infrastructure, and increasingly trusted AI systems all became viable attack paths.

For defenders, the challenge is no longer simply determining whether an action is allowed; it is determining whether that action makes sense in its wider context. Authentication, reputation, and provenance remain important, but they are no longer sufficient on their own. As attackers increasingly operate within trusted systems, the strongest signal is often a behavioral deviation: identifying when trusted activity no longer aligns with expected behavior.

‍

Credit to Nathaniel Jones (SVP, Global Threat Intelligence), Emma Foulger (Global Threat Research Operations Lead), Justin Torres (Senior Cyber Analyst), Daniel Levy (Threat Hunting Data Scientist)

‍
Edited by Ryan Traill (Content Manager)

‍

Appendix 1: Threat Research Methodology

Darktrace’s Threat Research team conducts extensive research across customer deployments to identify active threats, pinpoint key Indicators of Compromise (IoCs), and provide relevant threat intelligence. This research leverages Darktrace’s anomaly-based detection and involves thorough analysis and contextualization by the Threat Research team. Detected threats are promptly reported to the relevant customer security teams. When a customer has Darktrace’s Autonomous Response technology enabled, these threats are swiftly mitigated to prevent escalation.

Between January 1 and June 30, 2026, Darktrace investigated a wide range of cyber threats across its customer base. Many were identified as campaign-like activities targeting multiple customers, where clusters of similar tactics, techniques, and procedures (TTPs) and IoCs were seen affecting a significant number of customers within a short timeframe.

Statistics related to email are derived from aggregated Darktrace / EMAIL data across all cloud-hosted customer deployments between January 1 and June 30, 2026. Standard data-quality filtering was applied to exclude anomalous observations prior to aggregation. Regional statistics are based on relevant subsets of this dataset.

‍

Appendix 2: Campaigns - Regional and Sector Trends

While the above broad themes defined the threat landscape over the last six months, campaign clustering across the Darktrace customer base revealed how they manifested differently across sectors, regions, and industries.

Darktrace’s Threat Research team investigates a range of threats affecting its customer base. Through this research, campaign-like clusters of activity have been identified, in which common tactics, techniques, and procedures (TTPs), as well as infrastructure, are observed impacting a significant number of customers within a short timeframe.

Sectors and industries are classified using the Standard Industrial Classification (SIC) system to ensure consistent categorization. While the sector and regional insights in this report reflect broader global trends, they are also influenced by the distribution of Darktrace's customer base. For example, Finance, Manufacturing, and Education are strongly represented among Darktrace customers, which may result in a higher number of observed cases in these sectors. This reflects customer distribution rather than necessarily indicating elevated sector-specific risk. Similarly, regional trends may be influenced by the geographic distribution of Darktrace customers.

‍

Analysis of campaign clusters identified by the Darktrace Threat Research team during the first half of 2026 revealed distinct regional trends.

• Europe, Middle East & Africa (EMEA) dominated with 60% of all campaign cluster cases targeting this region.
• The Americas (AMS) was the next most affected region, with 30% of campaign cluster cases.
• The Asia-Pacific and Japan (APJ) region was less affected by campaign clusters, potentially indicating that threat actors placed a lower priority on the region and instead focused their efforts elsewhere.

Sector targeting also varied considerably by region:

• In EMEA, the Information and Communication was the most affected by a significant margin, representing 25% of all cases.
• In contrast, AMS targeting was more evenly distributed, with the Education, Public administration and defence, and Financial Insurance activities sectors all forming over 20% of AMS regional cases.
• Across APJ, campaign activity was spread more equally, with no single sector emerging as a dominant target.

‍

Several countries also stood out within their respective regions:

• The United States accounted for 60% of all campaign clusters within AMS.
• Japan represented 40% of campaign customer cases across APJ.
• In EMEA, the United Kingdom and Zimbabwe each accounted for 23% of identified cases, both being involved in a variety of campaign types.

‍

Inside the SOC & Threat Research 2026 Monthly Progression: From Access to Impact

Appendix 3: Bibliography

External

[1] https://www.darkreading.com/cyberattacks-data-breaches/jadepuffer-first-complete-llm-driven-ransomware-attack

[2] https://www.trendmicro.com/en_us/research/26/c/pawn-storm-targets-govt-infra.html

Darktrace Reading

1.        https://www.darktrace.com/blog/ai-llm-generated-malware-used-to-exploit-react2shell

2.        https://www.darktrace.com/blog/2025-cyber-threat-landscape-darktraces-mid-year-review

3.        https://www.darktrace.com/resources/annual-threat-report-2026

4.        https://www.darktrace.com/blog/when-trust-becomes-the-attack-surface-supply-chain-attacks-in-an-era-of-automation-and-implicit-trust

5.        https://www.darktrace.com/blog/hola-vpn-abuse-from-proxy-traffic-to-malware-and-cryptomining

6.        https://www.darktrace.com/blog/security-after-signatures-operating-in-a-world-of-pre-cve-disclosure-exploitation-collapsed-trust-boundaries-and-autonomous-systems

7.        https://www.darktrace.com/blog/when-ai-infrastructure-becomes-part-of-the-attack-surface

8.        https://www.darktrace.com/blog/cve-2026-1731-how-darktrace-sees-the-beyondtrust-exploitation-wave-unfolding

9.        https://www.darktrace.com/resource/understanding-chinese-nexus-cyber-tradecraft

10.   https://www.darktrace.com/blog/chinese-apt-campaign-targets-entities-with-updated-fdmtp-backdoor

11.  https://www.darktrace.com/blog/inside-zionsiphon-darktraces-analysis-of-ot-malware-targeting-israeli-water-systems

12.  https://www.darktrace.com/resources/the-state-of-cybersecurity-in-the-finance-sector

13.  https://www.darktrace.com/blog/from-click-to-command-behavioral-detection-of-applescript-led-macos-intrusions

14.  https://www.darktrace.com/blog/the-state-of-cybersecurity-in-the-finance-sector-six-trends-to-watch