Ongoing Purported AI-Assisted Identity Fraud Enables Unauthorized Access to Western Companies by North Korean IT Workers

North Korean operatives have reportedly used AI-generated identities to secure remote jobs or impersonate employers in order to infiltrate companies. These tactics allegedly support sanctions evasion through wage theft, credential exfiltration, and malware deployment. Workers reportedly use fake resumes, VPNs, and face-altering tools; some deploy malware like OtterCookie after embedding, while others lure targets via spoofed job interviews. AI systems are reportedly used to generate fake resumes, alter profile photos, and assist in real-time responses during video interviews.

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.3 Fraud, scams, and targeted manipulation

The incident involves North Korean operatives using AI-generated deepfakes and face-changing software to commit employment fraud and extort victim companies for financial gain.

Additional risk subdomains

• 2.1 Compromise of privacy by obtaining, leaking or correctly inferring sensitive information: Operatives compromised the personally identifiable information of dozens of U.S. citizens and exfiltrated proprietary corporate data to conduct extortion.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The incident was driven by human malicious actors intentionally using deployed AI tools to defraud and extort Western companies.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Risk Level 3: Limited Risk. The incident involves the use of AI-generated content, specifically deepfakes and face-changing software, to impersonate individuals during video calls and interviews.

AI system and alleged parties

• AI system: AI-generated profile pictures, ChatGPT (OpenAI)
• AI purpose: Deepfake Video Generation; Writing Assistant
• Behaviour type: Assistant
• Alleged developer: Unknown large language model developers, Unknown deepfake technology developers, OpenAI
• Alleged deployer: Yang Di, WaterPlum, Wagemole, Void Dokkaebi, UNC5267, Son Un Chol, Sok Kwang Hyok, Sim Hyon-Sop, Rim Un Chol, Ri Kyong Sik, Reconnaissance General Bureau, PurpleBravo, North Korean threat actors, Minh Phuong Ngoc Vong, Matthew Isaac Knoot, Lazarus Group, Ko Chung Sok, Kim Ye Won, Kim Sang Man, Kim Ryu Song, Kim Mu Rim, Jong Song Hwa, Jong Kyong Chol, Jang Chol Myong, Hyon Chol Song, Gwisin Gang, Government of North Korea, Famous Chollima, Department 53, Contagious Interview, Christina Chapman, Choe Jong Yong, Cho Chung Pom
• Alleged harmed parties: Western companies, Web3, Truth, SSA, Social Security Administration, Recruitment teams, Oleksandr Didenko, National security and intelligence stakeholders, macOS users, Jiho Han, IRS, Interviewees, Internal Revenue Service, Human resources staff, Hiring managers, Haoran Xu, Epistemic integrity, Employers, Developers, Cryptocurrency platforms, Companies in the United States, Chunji Jin, Blockchain projects, Andrew M.

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Substantial. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Substantial, indirect Minor
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Minor, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Minor, indirect Minor
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: The reports describe massive financial losses, including $6.8 million generated by one cell, $88 million generated by another, $866,255 from a third, and $500,000 in remediation costs for one company.

Directly caused: The scheme directly funneled millions of dollars to North Korea, including $88 million from one conspiracy, $6.8 million from another, and $866,255 from a third, representing illicit wages paid to fraudulent workers.

Indirectly caused: Victim companies incurred significant indirect costs, such as over $500,000 in auditing and remediation costs for one company, and hundreds of thousands of dollars in damages from extortion.

Inferred additional harm: Total global financial losses, including stolen wages, extortion payments, and cybersecurity remediation across the 300+ victim companies, are likely to exceed $100 million.

Malicious content

Reported: The reports describe the creation and use of AI-generated deepfakes and face-changing software to deceive employers during video calls.

Directly caused: Operatives created AI-generated deepfakes of company executives for Zoom calls and used face-changing software to impersonate U.S. citizens during job interviews.

Inferred additional harm: It is likely that numerous other deepfake videos and synthetic profile photos were generated to maintain the fake personas across various freelance and job platforms.

Privacy

Reported: The reports explicitly describe the theft of personally identifiable information (PII) of U.S. citizens and the exfiltration of proprietary corporate data.

Directly caused: Operatives stole and compromised the identities of more than 60 U.S. citizens, including using stolen U.S. passports and creating false tax liabilities.

Indirectly caused: Operatives exfiltrated sensitive corporate data, including proprietary source code, which was then used to extort victim companies.

Inferred additional harm: It is highly likely that the PII of thousands of other individuals was compromised or exposed during the infiltration of over 300 companies.

Epistemic

Reported: The reports describe the fabrication of identities, resumes, and online personas to deceive companies.

Directly caused: Operatives fabricated entire professional histories, online portfolios, and fake company websites to deceive employers into believing they were hiring legitimate U.S.-based workers.

Inferred additional harm: The widespread use of synthetic identities and fake credentials erodes trust in remote hiring ecosystems and online professional platforms.

People affected

• Occurrences reported: 1
• People reportedly harmed: 60
• People reportedly exposed: 10000

Potential causes

Management

• Prioritizing Speed Over Vetting: Management focused on rapid hiring of contract workers over security.
• Underestimating Insider Threats: Firms failed to assess risks of remote workers gaining privileged access.
• Inadequate Detection Budgets: Lack of investment in insider threat detection and monitoring software.

Technology

• AI-Generated Executive Deepfakes: Used in fake Zoom meetings to deceive employees and build false trust.
• Real-Time Face-Changing Software: Used during video interviews to bypass visual identity verification.
• AI Assistants for Interviews: Helped fraudulent candidates answer technical questions in real-time.

Data Inputs

• Stolen Personally Identifiable Info: Stolen PII of U.S. citizens used to create fake profiles.
• AI-Altered Profile Images: Photos edited to create realistic but fake developer personas.
• Fake Resumes and Portfolios: Resumes with fabricated experience and fake references submitted in volume.

Human Factors

• Overreliance on Video Verification: HR assumed video calls were secure, failing to detect deepfakes.
• Willingness of Domestic Facilitators: U.S. accomplices hosted laptop farms for financial gain.
• Lack of Skepticism in Interviews: Interviewers missed signs of prompt-taking and long pauses.

Process and Methods

• Inadequate Remote Vetting Methods: Hiring processes failed to thoroughly verify candidate identities.
• Weak Remote Access Controls: Lack of detection for unauthorized remote desktop tools on corporate laptops.
• No Ongoing Identity Verification: Companies failed to verify that the hired person was doing the work.

Regulatory Environment

• Sanctions Evasion Exploits: DPRK actors exploited gaps in freelance platforms to bypass sanctions.
• Lack of Platform Identity Checks: Freelance and job search platforms lacked robust identity verification.

Information quality

• Classification confidence: High
• Reason for confidence: The reports from the DOJ, Treasury, and major cybersecurity firms provide highly detailed, consistent, and corroborated accounts of the North Korean IT worker scheme, including specific indictment details, financial figures, and technical methodologies.
• Ambiguities identified: The specific AI models or platforms used to generate the deepfakes and assist with real-time interview answers are not named.
• Alternative interpretations: None. The evidence of coordinated state-sponsored employment fraud and extortion is overwhelming.

North Korean state-sponsored actors utilized AI-generated deepfakes and face-changing software to infiltrate over 300 Western companies, including government agencies and banks. By generating tens of millions of dollars to fund weapons programs and compromising proprietary data, this campaign represents a severe threat to economic and technological security.

• Overall national security impact: Severe
• Scope: Multiple nations
• Primary target: United States
• Other affected: United Kingdom and Europe
• Alleged perpetrator: North Korea

Threat characteristics

• Imminence: Long-term. Represents an ongoing strategic funding mechanism and cyber-espionage capability rather than an immediate tactical crisis.
• Autonomy: Human-controlled. AI tools like deepfakes and writing assistants were directly operated by human threat actors to augment their deceptive capabilities.
• Novelty: Evolved capability. Represents a significant advancement of existing employment fraud techniques through the integration of real-time AI face-changing and deepfake software.

Impact by dimension

• Physical security: Minor. No direct kinetic impact or critical infrastructure failures were reported, though unauthorized access to corporate networks poses potential latent risks.
• Information security: Substantial. State-sponsored threat actors used AI deepfakes to bypass identity verification, allowing infiltration of corporate networks and exfiltration of proprietary data.
• Sovereignty: Substantial. North Korean operatives successfully targeted government agencies and used illicit funds to bypass international sanctions and support state weapons programs.
• Economic security: Severe. Operatives funneled tens of millions of dollars to North Korea to fund weapons programs and exfiltrated proprietary source code from over 300 companies, requiring coordinated federal response.
• Societal stability: Minor. Identity theft of over 60 citizens caused personal harm and financial distress, but did not threaten large-scale social cohesion or human rights.