Important considerations to keep your IBM i secure in an AI-driven world.

By Pauline Brazil Ayala

No fewer than a dozen emails land in my inbox each day touting the latest use, benefit, or risk of AI. While artificial intelligence has become the technology industry’s buzzword of choice, dismissing it as a passing trend would be a mistake. AI is rapidly reshaping how technology is developed, deployed, and managed—and it is also reshaping the cybersecurity landscape.

For cybercriminals, AI is proving to be a powerful force multiplier. It enables attackers to create more convincing phishing campaigns, automate reconnaissance, develop malware more efficiently, and launch attacks at a scale previously difficult to achieve. As a result, organizations of all sizes face a threat environment that is faster, more adaptive, and harder to defend against.

When speaking with organizations that rely on IBM i, I frequently hear questions like, "How do AI-driven threats affect IBM i environments? Why should IBM i administrators and security professionals be concerned? And what practical steps should organizations take to reduce their risk?”

In this article, we’ll examine how AI is reshaping the cybersecurity landscape, why these changes matter even for organizations not actively using AI, and what IBM i professionals should be doing today to prepare for emerging threats.

The New Wave of Emerging Attacks

Cybersecurity has been a critical concern for Fortune 500 and Global 2000 organizations for decades. In recent years, however, the threat landscape has expanded well beyond large enterprises. Small and medium-sized businesses are increasingly in the crosshairs of cybercriminals, driven by the expansion of hacker groups worldwide and the rise of “Ransomware-as-a-Service” offerings that have lowered the barrier to entry.

Launching a sophisticated cyberattack used to require significant expertise, resources, and time. Today, many of those barriers have been reduced or removed entirely. Criminal ecosystems now resemble commercial software markets, where attackers can purchase ransomware kits, phishing templates, and access to compromised systems through underground channels.

Artificial intelligence has accelerated this evolution. CrowdStrike reported in its 2026 Global Threat Report that in 2025, AI-enabled adversaries increased attacks by 89% year over year.  AI acts as a force multiplier, automating tasks that once required skilled operators. Phishing emails can now be generated with near-perfect grammar and tailored messaging. Social engineering campaigns can be scaled and personalized in ways that were previously impractical.

AI is also improving reconnaissance, vulnerability discovery, and aspects of malware development. While it does not eliminate the need for human attackers, it significantly increases their productivity. The result is a higher volume of attacks that are more targeted, more convincing, and more difficult to detect.

AI-Enabled Phishing Attacks

One of the most immediate impacts of AI is the transformation of phishing. AI tools now generate highly polished, context-aware messages that closely resemble legitimate business communications.

In the past, phishing attempts were often identifiable due to poor grammar or inconsistent formatting. Those indicators have largely disappeared. Messages can now be tailored using publicly available information about an organization, its employees, and its business relationships.

This has led to a new class of spear-phishing attacks that are significantly more convincing and more difficult to detect. As a result, employees can no longer rely on informal cues to distinguish legitimate communication from malicious activity.

When a phishing attack succeeds, the attacker gains full access to the identity and privileges of the compromised user. At that point, like any connected system, IBM i can also become a target. The Integrated File System (IFS) requires particular attention because improperly secured IFS objects and excessive access permissions can provide attackers with opportunities to encrypt or damage critical data.

Deepfakes

Deepfake technology represents another major evolution in AI-enabled cybercrime, extending social engineering beyond text-based communication into voice and video impersonation. Using AI-generated audio or video, attackers can convincingly imitate executives, managers, vendors, or trusted partners.

Voice cloning, in particular, has already been used in real-world incidents in which employees were tricked into authorizing financial transfers or disclosing sensitive information. These attacks are often time-sensitive and rely on authority pressure—an employee receives what appears to be a legitimate urgent request from a senior leader and acts without fully verifying authenticity.

AI-Enabled Vulnerability Exploitation

One of the most significant cybersecurity implications of AI is not just its ability to discover vulnerabilities, but its ability to quickly turn known vulnerabilities into working exploits.

Traditionally, there has been a window of opportunity between the release of a security patch and the widespread availability of exploit code targeting the underlying vulnerability. Security teams often relied on this "patch gap" to evaluate, test, and deploy updates before attackers could effectively weaponize the flaw. Now, AI is rapidly shrinking that window.

Recent research has demonstrated just how quickly this shift is occurring. Anthropic's Claude Mythos model was shown to generate working exploits for known vulnerabilities in a matter of hours rather than the days or weeks that exploit development has historically required. In testing against recently disclosed Firefox and Windows vulnerabilities, the model produced proof-of-concept and working exploit code at a pace that significantly compressed the traditional timeline between vulnerability disclosure and exploitation.

The broader implication is that threat actors no longer need the same level of reverse-engineering expertise once required to weaponize newly disclosed vulnerabilities. AI can assist in analyzing patches, identifying the underlying security flaw, and generating exploit code much faster than human researchers working alone. Security researchers have begun referring to this phenomenon as the transition from "N-days" to "N-hours," where the time required to weaponize a disclosed vulnerability is measured in hours rather than days.

As AI continues to accelerate the exploitation of vulnerabilities, organizations will need to rethink traditional patch management timelines and remediation processes. Security teams can no longer assume they have days or weeks to address critical vulnerabilities. The race between defenders and attackers is becoming increasingly compressed, and AI is helping attackers move faster than ever before.

For IBM i environments, this means maintaining awareness of vulnerabilities not only in the IBM i operating system but also in the software components running on the platform, including third-party applications, open-source libraries, web services, APIs, and other integrated technologies. Regularly reviewing IBM i Security Bulletins and reported vulnerabilities through sources such as the Common Vulnerabilities and Exposures (CVE) database can help organizations identify issues that may require attention and prioritize appropriate remediation efforts.

AI-Generated Code Exploitation

AI is also changing how malicious code is created and adapted. While AI-assisted development improves productivity for legitimate users, it also lowers the barrier for attackers.

Tasks that once required programming expertise—such as writing malware, automation scripts, or credential-harvesting tools—can now be generated through natural language prompts and iterated rapidly.

This enables a broader range of threat actors to participate in cybercrime. Attackers can quickly generate tool variations, increasing their ability to evade detection and scale operations.

Additionally, as developers increasingly rely on large language models (LLMs) to generate code, there is a growing risk that vulnerabilities introduced in AI-generated output will go unnoticed and be exploited. Because the code is produced by a trusted tool, there is a tendency for developers to become less rigorous in applying secure design principles, conducting thorough code reviews, and performing threat modeling, which can lead to a false sense of confidence in the output. Compounding the issue, attackers can use the same tools to generate similar code paths and systematically analyze them for inherited weaknesses to exploit.

AI-Enhanced Reconnaissance

Reconnaissance is the foundation of most cyberattacks, and AI has made it faster and more comprehensive.

AI systems can analyze large volumes of publicly available data—such as corporate websites, job postings, and social media—to build detailed profiles of an organization’s infrastructure and personnel.

Attackers can also correlate disparate data points to build a clearer picture of an organization’s technology landscape, including likely platforms, integration points, and potential access pathways. Information about users with elevated privileges, such as IBM i system administrators, becomes especially valuable because it can reveal who has the authority to access critical systems and data. The more attackers can learn about privileged users, applications, and data flows before an attack begins, the more effectively they can tailor their approach.

This significantly reduces the time required to move from selecting a target to actionable intelligence - the information that can be directly used to support an attack.

Securing IBM i in the Age of AI

Modern IBM i environments rarely consist solely of traditional IBM i processes. Most organizations now operate a blend of native IBM i applications, Open Source software running in PASE, web services, and third-party solutions. Each layer expands functionality, but also introduces additional attack surface that must be actively maintained, monitored, and secured.

IBM i environments can be compromised either through vulnerabilities on the platform itself or through connected systems and users. While organizations should remain diligent about patching and securing IBM i, they must also recognize that stolen credentials, phishing attacks, vulnerable endpoints, and weaknesses in integrated applications can provide attackers with a path to critical IBM i data and operations.

There is also an important shift occurring in attacker capability: access to IBM i expertise is becoming less of a barrier than it once was. IBM i knowledge has typically been relatively specialized, and attackers often focused on more commonly understood platforms such as Windows or Linux. However, AI tools now make it significantly easier to understand unfamiliar systems, interpret IBM i concepts, and generate actionable guidance for interacting with the platform. Tasks such as navigating the operating system, understanding security authorities, querying system data, or scripting administrative actions are no longer limited to experienced IBM i professionals. With this knowledge barrier reduced, the likelihood increases that IBM i will be more deliberately targeted or more effectively exploited when access is gained.

The reality is that AI is increasing both the speed and effectiveness of cyberattacks. To reduce risk, organizations should adopt a Zero Trust approach to IBM i security. Rather than relying on a single layer of protection, Zero Trust assumes that any user, device, or connection could be compromised and requires continuous validation and access restrictions throughout the environment.

Several layers can be implemented:

Network Security

IBM i data and services can be accessed through numerous interfaces, including FTP, ODBC, JDBC, SQL services, Remote Command, Telnet, and web-based applications. When these services are enabled without proper controls, they create additional pathways into the system.

Organizations should implement IBM i-specific controls such as exit programs, network access controls, and socket-level monitoring to govern who can access resources, from where, and through which interfaces. These controls provide an important layer of protection beyond corporate firewalls, VPNs, and MFA solutions. (Read more on securing network traffic here.)

Object Authorities and IFS Permissions

Object-level security remains one of the most important components of a Zero Trust strategy. Too often, sensitive libraries, files, directories, and IFS objects are left overly accessible, with organizations relying solely on application security to enforce restrictions.

Access controls should be implemented at the object, library, file, and directory levels to ensure that data remains protected regardless of how a user accesses the system. For example, a file may be protected within a web application, but if the underlying object authority is configured as *PUBLIC *ALL, a user connecting through FTP or SQL may be able to bypass application controls entirely. Proper authority management helps ensure that data protection remains consistent across all access methods. (Learn more about taking a zero-trust approach to IFS security here.)

Privileged Access Management

Special authorities should be granted only when required and reviewed regularly. Administrative capabilities such as *ALLOBJ, *SECADM, and other elevated privileges significantly increase risk if compromised.

Organizations should establish formal processes for approving, reviewing, and monitoring privileged access to ensure that elevated authority is limited to users with legitimate business requirements.

Multi-Factor Authentication

Multi-factor authentication should be implemented wherever possible, particularly for administrative accounts and remote access services. As AI-driven phishing and social engineering attacks become increasingly convincing, passwords alone no longer provide sufficient protection.

Extending MFA to IBM i-specific access methods, including remote server interfaces and administrative connections, can significantly reduce the likelihood of unauthorized access through compromised credentials.

Least-Privilege Access

Users should be granted only the minimum level of authority necessary to perform their job functions. Excessive privilege remains a common weakness in many IBM i environments and can dramatically increase the impact of a compromised account.

Organizations should regularly review user authorities, remove unnecessary special authorities, and closely monitor activity associated with highly privileged profiles. (More on this topic here.)

Security Configuration

Strong security begins with proper system configuration. IBM i security-related system values, network settings, auditing controls, and authentication policies should be reviewed regularly to ensure they align with current security best practices and organizational requirements.

Misconfigurations often create opportunities for attackers that are far easier to exploit than software vulnerabilities. (Read more here about if your IBM i is configured securely.)

Vulnerability Management

Vulnerability management must remain a continuous process. Organizations should maintain current IBM i releases, apply security-related PTFs promptly, and ensure that Open-Source software and third-party applications are kept up to date.

The IBM i platform is not immune to vulnerabilities, and in recent years, there has been an increase in reported CVEs affecting both the operating system and supporting components. Maintaining a disciplined patching and update strategy is essential to reducing exposure and minimizing opportunities for exploitation. (Learn more about CVEs here.)

Each additional layer of security strengthens a Zero Trust implementation. While no individual control can eliminate risk, combining strong configuration management, access controls, monitoring, authentication, and vulnerability management creates a significantly more resilient IBM i environment against increasingly capable AI-driven threats.

The Bottom Line

Artificial intelligence is not a passing trend—it is here to stay and is rapidly evolving the technology landscape, for honest employees and cybercriminals alike. As AI capabilities continue to advance, so will the sophistication, speed, and scale of AI-enabled cyberattacks.

For organizations that depend on IBM i to run mission-critical business processes, this reality cannot be ignored. Now more than ever, organizations must take a proactive approach to security by regularly reviewing configurations, implementing strong access controls, monitoring for suspicious activity, and ensuring their IBM i environments align with current security best practices. The threats are changing, and our security strategies must evolve with them.

References

• CrowdStrike Report: https://go.crowdstrike.com/2026-global-threat-report.html
• Mythos Example: https://www.securityweek.com/claude-mythos-turns-n-days-into-n-hours-with-rapid-exploit-creation/