Defending Against Native Phishing: How AI and Cloud Tools Are Changing Cybersecurity

Attackers are upping their game in the world of phishing, combining the power of artificial intelligence and native cloud tools to build attacks that are nearly indistinguishable from legitimate IT workflows. The latest trend, “native phishing,” leverages trusted Microsoft 365 (M365) infrastructure, AI-powered no-code web builders, and advanced social engineering to subvert conventional security controls and exploit user trust. Both the sophistication and success rate of these attacks should be an urgent wake-up call for IT, security, and business leaders.

What Makes Native Phishing So Devastating?​

Traditional phishing campaigns—those that rely on spoofed external addresses and obviously malicious attachments—are becoming less effective. Advanced spam filters, improved user training, and robust technical policies have dramatically reduced the odds that a generic, external phishing email will slip through. But as defenses have evolved, so have attackers.
Native phishing flips the traditional model on its head: rather than delivering malicious content from outside the organization, attackers turn “trusted” internal applications into attack vehicles. After compromising a single Microsoft 365 account (through credential stuffing, password reuse, weak authentication, or a successful prior phishing campaign), the threat actor can weaponize legitimate tools like OneDrive, SharePoint, and OneNote against the company from within its own perimeter.

A Real-World Example: Weaponizing OneNote and OneDrive​

In incidents documented by Varonis Threat Labs, attackers demonstrated an alarming level of creativity and technical acumen. Here’s how one such attack unfolds:

• Initial Compromise: The attacker gains control of a company user’s Microsoft 365 account.
• Payload Creation: Using the compromised account, they craft a malicious OneNote file—a tool not protected by Microsoft’s “Protected View” feature, making it ideal for deception.
• Internal Distribution: This OneNote file is uploaded into OneDrive, and the sharing functionality is used to send a genuine Microsoft notification to hundreds of internal users. Since the sharing notification is an automated Microsoft alert, it appears wholly legitimate to the average recipient, often slipping past traditional scans for suspicious attachments or spoofed headers.
• Credential Harvesting: The OneNote file itself contains links to a lookalike login page—a phishing site that closely mimics the company’s actual authentication portal.
• User Deception: Employees, trusting the internal source and Microsoft branding, are much more likely to click and enter their credentials, believing it to be a routine access request.

This chain of events uniquely capitalizes on the architecture and trust relationships of Microsoft 365. It not only makes detection far more challenging, but also increases the success rate of phishing attempts. The psychology at play is deeply effective: users are trained to question odd requests from outside the organization, but rarely question notifications from Microsoft relating to documents shared by colleagues.

Why OneNote is the Perfect Vehicle​

The attackers’ choice of OneNote is no accident. Microsoft’s Protected View provides an extra security layer when opening documents from unfamiliar sources (like email). But OneNote files escape this scrutiny. Their flexible layout allows attackers to create visually convincing fake login prompts, embed clickable links, and use formatting to direct attention exactly where they want it. Combined with the internal source and Microsoft-branded sharing notification, the attack blends seamlessly into normal business operations.

The AI and No-Code Phishing Revolution​

What happens after the user clicks the link in the malicious OneNote? This is where AI-powered no-code platforms enter the picture.
Platforms like Flazio, ClickFunnels, and JotForm are designed, legitimately, to let organizations build beautiful web pages and complex forms without any coding knowledge. However, these same strengths make them a powerful tool for attackers. In Varonis’s research, attackers used Flazio to clone company login pages with pixel-perfect detail. The fake login portals were hosted on otherwise harmless URLs provided by these services, often with SSL certificates, making the phishing attempt look even more authentic.

The Role of AI​

Artificial intelligence, increasingly baked into these platforms, streamlines the process even further. With AI-assisted design, attackers can automate site generation—copying branding, layouts, and language with minimal effort. AI tools can rapidly test variations, maximize conversion (credential theft) rates, and even generate dynamic responses to user input in real time. The bar for technical sophistication lowers with every AI-powered update, allowing less skilled cybercriminals to execute advanced, high-fidelity attacks at scale.

Defeating Native Phishing: Why Traditional Defenses Struggle​

The underlying danger of native phishing is that it fundamentally breaks the implicit trust organizations place in their own systems and notifications. Conventional defense tools—such as external sender warnings, attachment scanning, and domain reputation scoring—are rendered less effective because:

• The emails and notifications are sent from legitimate company or Microsoft domains.
• Standard attachment and URL scanning is rarely triggered with internal sharing links.
• The payload (malicious OneNote or embedded link) is hosted on platforms already designated as safe in most allow-lists.
• Microsoft 365’s sharing capabilities, by design, facilitate rapid internal collaboration. Unless strictly configured, it is trivial for a single compromised user to share content widely within the organization.

Corporate anti-phishing training, meanwhile, often focuses on spotting the external attacker: the clumsy grammar, the odd sender address, the unexpected LinkedIn connection. But native phishing is a betrayal of perceived safety—from a colleague, on Microsoft letterhead, referencing actual internal projects.

How Organizations Can Fight Back​

Fortunately, all is not lost. Varonis and cybersecurity experts recommend a set of overlapping safeguards, blending technical controls with continuous user education:

1. Enforce Multi-Factor Authentication (MFA) and Conditional Access​

• MFA is Critical: If all accounts use MFA, the odds of an initial credential compromise drop dramatically—even if a password is stolen, access is denied without a second authentication factor.
• Conditional Access Policies: Group- and location-based access controls can add another layer, for example, denying logins from unusual countries or restricting access when risky behavior is detected.
• Zero Trust Principles: Enforcing least privilege, device hygiene, and ongoing authentication reduces the blast radius of any single compromised account.

2. Run Regular Phishing Simulations​

• Realistic Scenarios: Simulations should not just focus on external attacks but replicate internal, native phishing—using Microsoft 365 tools, authentic-looking sharing notifications, and familiar file types.
• User Reporting Metrics: Track improvements in employee detection and reporting, not just simulated “click rates.”

3. Review and Harden Microsoft 365 Sharing Policies​

• Restrict Sharing: Limit OneDrive, SharePoint, and OneNote sharing to only those with a genuine need. Disable external sharing unless absolutely necessary.
• Conditional Sharing Alerts: Trigger alerts whenever sensitive files or new sharing patterns are detected—such as a user suddenly sharing to hundreds of colleagues.
• Audit Logs: Routinely review sharing and access logs for abnormal behaviors.

4. Build Clear Internal Reporting Channels​

• Frictionless Reporting: Employees should know exactly how and where to report suspicious activity, with minimal barriers. Fast, visible action in response encourages future vigilance.
• Security Champion Programs: Appoint team-based security advocates to promote awareness and funnel local observations back to IT.

5. Monitor and Alert on No-Code Platform Usage​

• Traffic Analysis: Monitor for traffic to and from common no-code web page builders like Flazio and ClickFunnels, especially when those domains appear in authentication workflows.
• Allow-List Management: Continuously review URL allow-lists to ensure emerging threats are not inadvertently sanctioned organization-wide.

The Double-Edged Sword of No-Code and AI​

No-code development and AI assistants are transforming business productivity, democratizing app creation, and unlocking new forms of innovation. But every productivity boost introduces a corresponding security challenge. Because attackers use the same SaaS features, web-facing services, and streamlined workflows as the rest of the business, defenders must outpace not just technical exploits, but also weaponized convenience.
The ability to rapidly build, scale, and personalize convincing phishing portals using legitimate software blurs the lines between normal business and malicious activity. As these platforms further incorporate generative AI, the speed, believability, and scope of attacks amplify. Attackers are no longer constrained by their own design skills or coding ability; they can iterate, localize, and adapt their lures in minutes.
Organizations must thus continually re-assess permissions for no-code tools and AI-driven services, integrating their use with security monitoring and centralized identity management. Where possible, implement API integration with your security operations center to receive alerts about suspicious account creation or platform use.

Insider Threats and Lateral Movement​

The risk from native phishing is not limited to credential theft. Once inside, attackers can:

• Exfiltrate sensitive documents.
• Launch additional attacks (malware, ransomware, business email compromise) from a trusted user’s account.
• Remediate or remove forensic evidence, making post-incident investigations difficult.
• Leapfrog across departments and systems, escalating privileges or gaining additional footholds.

The attacker’s ability to operate with an internal persona allows them to mask their movements and mimic a legitimate user. For example, if a threat actor with compromised credentials is running new sharing campaigns, accessing atypical resources, or triggering internal policy exceptions, behavioral analytics and advanced security information and event management (SIEM) systems become vital detection tools.

The Human Factor: Social Engineering Remains Central​

Even with all technical controls in place, attackers will continue to refine their social engineering. Internal phishing attacks are so effective because users are conditioned to trust:

• Colleagues and known contacts.
• Brand recognition (Microsoft, company logos).
• Standardized workflow communications.

Attack simulations, targeted training, and relentless reinforcement of a “pause and verify” culture are vital. For IT administrators, regular reviews of access logs, privilege assignments, and anomalous workflow behaviors can make the difference between a minor incident and a large-scale data breach.

The Microsoft 365 Ecosystem and Its Discontents​

Microsoft has worked tirelessly to evolve security features in its M365 suite, but the platform’s openness and focus on collaboration can be a double-edged sword. Security features such as Safe Links, anti-phishing policies, and cloud app security provide valuable protection, but require skillful configuration and continuous maintenance. The persistent threat of internal abuse—through intentionally malicious insiders or through compromised accounts—will always challenge even the most robust security posture.
Greater focus on adaptive access controls, integration of behavioral analytics, and alignment with industry-recognized standards such as NIST and CIS can help organizations strengthen their defense-in-depth strategies. However, these measures must be tailored to each organization’s unique collaboration patterns and risk tolerance.

Risks of Over-Reliance on Automation​

AI-driven automation, while powerful, is not infallible. Over-reliance on automated threat detection can result in missed subtle attacks that blend with routine business. A hybrid approach—combining machine learning with human-in-the-loop review—yields much better outcomes, especially for identifying low-volume, high-consequence phishing events staged from internal accounts.
Furthermore, as generative AI becomes a common tool for both attackers and defenders, the race to stay ahead becomes more frantic. Organizations should ensure transparency and auditability in their own AI-powered security infrastructure, lest they become blind to creative new attack methodologies.

Conclusion and a Forward-Looking Perspective​

Native phishing campaigns demonstrate how cloud adoption, no-code technologies, and AI can unintentionally empower cybercriminals. By infiltrating Microsoft 365 environments and exploiting trusted communication channels and automation, attackers achieve a level of dexterity and authenticity that complicates detection and response.
To combat this evolving threat landscape, every organization must:

• Treat all internal collaboration tools as potential attack vectors.
• Take a holistic approach to zero trust, identity-first security.
• Continuously educate users with cutting-edge, realistic simulations.
• Harness both technical innovation and the vigilance of the human workforce.

The convergence of AI, no-code development, and cloud-native workflows will only accelerate, bringing both business opportunity and escalating risk. Those who adapt their defenses, close the “trust gaps,” and encourage ongoing vigilance will be best positioned to defend against the next generation of phishing—and whatever comes after.
For IT and security teams, this is not merely a technical challenge, but an organizational imperative. The tools, skills, and habits you invest in today will define your ability to keep tomorrow’s threats at bay. As native phishing demonstrates, the battlefield is already inside the gates; it’s up to us to fortify them.

---

Source: Dataconomy Attackers use “native phishing” with M365 and AI tools