How AI-Powered Attacks Target Defense Contractors

Defense contractors have always been prime targets for nation-state cyber espionage—but 2025 and 2026 have marked a measurable shift in how those attacks are conducted. Adversaries from China, Russia, Iran, and North Korea are now actively integrating artificial intelligence into their attack operations, using AI tools to accelerate reconnaissance, generate hyper-personalized phishing lures, automate vulnerability discovery, and maintain persistent access inside contractor networks. The result is a threat environment where even a 12-person defense contractor with active DoD contracts can find themselves the subject of a sophisticated, targeted intrusion campaign. If your company touches Controlled Unclassified Information (CUI), you are a target—and your adversaries are now better equipped than ever.

Which Nation-State Groups Are Targeting the Defense Industrial Base?

The Defense Industrial Base faces persistent targeting from multiple well-resourced, state-sponsored advanced persistent threat (APT) groups. Each operates with distinct objectives and methods:

• China-nexus actors (Volt Typhoon, Salt Typhoon, Flax Typhoon): Chinese APT groups have been the most aggressive in targeting DIB organizations in recent years. Their campaigns focus on exfiltrating controlled technical information, intellectual property, export-controlled specifications (ITAR/EAR), and personnel data from defense programs. China-nexus actors have become known for exploiting zero-day vulnerabilities in edge devices—VPNs, routers, and firewalls—to gain initial access while evading endpoint detection. Once inside, they deploy multiple malware families and maintain long-term persistent access to high-value networks, including managed service providers that serve defense contractors.
• Russia-nexus actors (APT28/GRU Unit 26165, Sandworm): Russian APT groups prioritize military, logistics, and defense technology targets. APT28 has been linked by a U.S.-led joint advisory to campaigns targeting dozens of logistics providers, IT service firms, and organizations supporting defense programs. Their intrusions rely heavily on spear phishing, credential theft, and lateral movement through trusted partner environments—including prime contractor networks accessed via subcontractor compromises.
• North Korea (Lazarus Group, Kimsuky): North Korean actors use job-themed lures, fake recruiter personas, and tailored phishing campaigns specifically targeting defense sector employees. Their campaigns often seek technology transfer, defense specifications, and financial access.
• Iran-nexus actors (UNC1549/Imperial Kitten): Iranian groups have been observed spoofing recruitment portals and resume builder applications to deliver malware to aerospace and defense personnel. Mandiant researchers documented ongoing espionage operations by UNC1549 targeting organizations in the aerospace, aviation, and defense industries, with the group gaining initial access through highly tailored phishing and supply chain compromises.

According to DC3 DCISE's CY2025 Q2 report, 12% of all mandatory cyber incident reports submitted to DC3 by DIB companies involved ransomware, underscoring that the threat isn't limited to espionage—criminal actors also actively target the defense supply chain.

How Are Adversaries Using AI to Attack Defense Contractors?

AI has fundamentally lowered the cost and increased the quality of nation-state cyber operations. Researchers at Google's Threat Intelligence Group (GTIG) confirmed in February 2026 that nation-state actors are now using AI tools—including large language models (LLMs)—as essential components of their attack chains. Specifically:

• AI-generated spear phishing: LLMs generate hyper-personalized, culturally nuanced phishing lures that mirror the professional tone of a target organization. Researchers at Harvard found that AI-enhanced spear phishing performs as well as or better than human-crafted attacks—at a fraction of the cost. One analysis estimated that automating the phishing attack chain with AI can reduce spear phishing costs by up to 99% at scale, eliminating the long-standing tradeoff between targeting quality and operational volume.
• Rapport-building multi-turn phishing: Rather than a single malicious email, AI models are being used to maintain multi-turn, believable conversations with targets—building trust over days or weeks before delivering a malicious payload. For defense contractors whose employees regularly communicate with contracting officers, auditors, and technical partners, this technique is especially difficult to detect.
• Automated reconnaissance: AI tools are used to scrape publicly available information about target organizations—LinkedIn profiles, contract award databases, government procurement portals, conference presentations—to build detailed target profiles and identify the highest-value individuals to impersonate or approach.
• Automated vulnerability discovery: AI-assisted tools accelerate the identification of exploitable vulnerabilities in a contractor's internet-facing infrastructure, reducing the reconnaissance-to-exploitation timeline significantly.
• ClickFix social engineering at scale: GTIG documented AI-hosted social engineering content—including on legitimate AI platforms like Gemini—designed to trick users into copying and executing malicious commands in their terminals. This technique, first observed in late 2025, represents a new vector that bypasses traditional email security controls.

What Does a Targeted Attack on a Defense Contractor Actually Look Like?

For small and mid-size defense contractors, the initial intrusion rarely looks like a sophisticated nation-state operation. It typically looks like this:

1. Reconnaissance: The adversary identifies your company through contract award notices, CAGE code lookups, or LinkedIn. They identify employees with clearances, technical roles, or financial authority.
2. Initial access via phishing or credential theft: A targeted employee receives a convincing email—perhaps appearing to be from a contracting officer, a teaming partner, or a technical reviewer. The AI-generated message has no spelling errors, references a real contract by name, and includes a plausible request (review a document, verify banking information, update portal credentials).
3. Persistence: Once credentials are obtained or malware is installed, the adversary establishes persistence through legitimate tools (living-off-the-land techniques), making detection difficult. They may maintain read-only access for months before taking any visible action.
4. CUI exfiltration: Technical drawings, specifications, test data, and program documents are staged and exfiltrated—often over encrypted channels that blend with legitimate traffic. By the time the contractor realizes they've been breached, the information has long since been collected.

A February 2025 case study of a 12-person, veteran-owned defense contractor found that 67% of employees had leaked credentials, the company had zero email authentication configured (no DMARC or DKIM), and their simulated red team showed a 92% probability of data exfiltration—all while the company held active CMMC obligations. This is not an outlier. It is the norm for underprepared small contractors across the DIB.

Which CMMC and NIST 800-171 Controls Address AI-Powered Threats?

CMMC 2.0 and NIST SP 800-171 contain a set of controls specifically relevant to defending against social engineering, credential attacks, and persistent intrusion—the core elements of AI-powered nation-state campaigns:

• AT.L2-3.2.1 and 3.2.2 (Awareness and Training): Security awareness training is mandatory under CMMC Level 2—and it is one of the controls that cannot be placed on a POA&M. Every user who accesses CUI must receive regular training. AI-powered phishing requires AI-powered training in response: traditional click-through awareness modules are insufficient against hyper-personalized lures.
• IA.L2-3.5.3 (Multi-Factor Authentication): MFA for all accounts with access to CUI is required and non-deferrable. Most AI-assisted phishing campaigns ultimately aim to harvest credentials—MFA is the most reliable barrier against credential-based initial access.
• AC.L2-3.1.13 (Remote Access): Requires the use of cryptographic mechanisms to protect confidentiality of remote access sessions. VPN implementations should be current and patched; edge device exploitation is a primary Chinese-nexus initial access vector.
• SI.L2-3.14.2, 3.14.4, 3.14.5 (Malware Protection): Requires malware protection at appropriate system locations, current signature updates, and periodic scanning. Behavioral detection is now essential—signature-based tools alone will not catch AI-crafted payloads or living-off-the-land techniques.
• CA.L2-3.12.1 (Security Assessment): Requires periodic assessment of security controls to verify they're operating effectively. Annual penetration testing focused on social engineering vectors—specifically AI-assisted phishing simulations—provides meaningful assurance.
• AU.L2-3.3.1 and 3.3.2 (Audit and Accountability): Comprehensive audit logging and review are foundational to detecting the kind of persistent, low-and-slow intrusions that nation-state actors conduct after initial access. Many small contractors have logging configured but no one reviewing it—an unmonitored log is as useful as no log at all.

What Does the DoD DIB Cybersecurity Program Offer to Help?

The DoD operates a voluntary DIB Cybersecurity Program through DC3 DCISE (Defense Cyber Crime Center, Defense Industrial Base Collaborative Information Sharing Environment). DIB companies can enroll to receive threat intelligence sharing, cyber analysis and diagnostics, and remediation consultation. DCISE publishes quarterly reports on DIB-specific threat trends—including threat actor TTPs, observed ransomware variants, and emerging attack vectors targeting defense contractor networks.

Participation in the DIB CS Program doesn't satisfy CMMC requirements, but the intelligence sharing can significantly improve your threat-informed security decisions. Given that nation-state actors maintain persistent access inside contractor networks for months before detection, having access to current adversary TTPs is a meaningful defensive advantage.

Additionally, CISA regularly publishes joint advisories with NSA and FBI specifically addressing nation-state threats to critical infrastructure and the defense sector. Contractors should subscribe to CISA's free Known Exploited Vulnerabilities (KEV) catalog and prioritize patching assets on that list—many of the edge device vulnerabilities actively exploited by China-nexus actors appear on the KEV list.

What Should Defense Contractors Do to Defend Against AI-Powered Attacks?

The practical defense posture for a small or mid-size defense contractor in 2025 and 2026 must address both the human and technical dimensions of AI-enhanced attacks:

• Deploy AI-aware phishing simulation and training: Traditional phishing simulation tools test against generic lures. Modern programs use AI-generated, context-aware simulations that reflect the quality of actual nation-state campaigns. Train your team to be suspicious of urgency, out-of-band requests, and any communication that asks for credentials or document access—even from known contacts.
• Enforce MFA everywhere, with phishing-resistant methods where possible: FIDO2-based authenticators (hardware security keys) are significantly more resistant to AI-assisted credential harvesting than SMS or app-based MFA. For cleared personnel and system administrators, phishing-resistant MFA should be the standard.
• Configure email authentication (DMARC, DKIM, SPF): These controls prevent your domain from being spoofed in AI-generated attacks targeting your partners, customers, and contracting officers—and help filter inbound spoofed messages at your mail gateway.
• Maintain current patching on all internet-facing assets: Edge devices are the primary initial access vector for Chinese-nexus actors. A documented vulnerability management program—with high-severity CVEs remediated within 30 days—is both a NIST 800-171 requirement (control 3.11.3) and a critical operational control.
• Implement 24/7 SIEM monitoring with threat intelligence integration: Detecting nation-state intrusion requires visibility into authentication events, lateral movement, data staging, and anomalous outbound traffic. Many small contractors lack the internal capacity for continuous monitoring—a managed SIEM or MDR service with defense sector threat intelligence can fill that gap.

See also our article on Ransomware Defense for DoD Contractors, as many nation-state initial access operations are later leveraged by criminal ransomware affiliates through credential marketplaces.

What Should Defense Contractors Do Next?

Nation-state actors have unlimited patience, significant technical resources, and now, AI-powered tools to help them identify and exploit the weakest contractors in the defense supply chain. The question isn't whether your company will be targeted—it's whether you'll be prepared when it happens.

NorthStar Technology Group works with defense contractors to build CMMC-aligned security programs that address the specific threat profile of the DIB: AI-enhanced phishing defenses, 24/7 threat monitoring, identity security, and the technical controls required for NIST SP 800-171 compliance. If you're concerned about your organization's exposure to nation-state threats or want to understand your current security posture against these attack vectors, reach out to our team at northstartechnologygroup.com/services/dod-cmmc.