
Deepfake Interview Scam: npm Malware Exposed
You aced the interview. They aced your wallet.
The Scheme
Fake recruiters conduct video interviews for high-paying crypto jobs. Camera stays off or uses AI-generated faces. The 'technical test' is a GitHub repo with wallet-draining malware hidden in node_modules.
Your dream job is their payday.
Camera off. Perfect offer. Technical test. npm install. Wallet drained. Congratulations — you got the job. Just not the one you applied for.
The Hard Truth
In 2025, deepfake technology became accessible to individual scammers. A camera that's 'broken' or a face that seems slightly off during a video call isn't a technical issue — it's a warning sign.
If an interviewer asks you to run code before you've verified their identity through official channels, you're not being evaluated. You're being targeted.
The '$180K Senior Developer' role that found you on LinkedIn? It exists only to find developers with crypto wallets.
The Anatomy of This Scam
This scam specifically targets developers and technical professionals who regularly receive legitimate recruiter messages. The interview feels real because it's designed to feel real. The salary is high because high salaries stop questions.
The malware isn't in the code you review. It's in the postinstall scripts you never check.
According to security researchers, npm supply chain attacks increased 300% in 2024. Scammers now combine social engineering with technical exploits — they don't just send you malware, they make you install it willingly.
Synthetic faces and fake recruiter messages are two forms of AI-powered crypto scams. Verify the identity through an independent channel before running the test.
The interview was the social engineering. The repo was the payload.
Scammer Vocabulary Decoded
They use familiar developer terms to make the scam feel routine. Here's what each actually means:
"Technical Assessment"
What it sounds like:
A coding test to evaluate your skills — standard in tech hiring. You solve problems, they review your code.
What it actually is:
A GitHub repo containing a hidden postinstall script. When you run npm install, it silently extracts browser extension data, including wallet credentials, and sends them to attacker servers.
"Camera Issues"
What it sounds like:
Bad internet connection, broken webcam — common technical problems during video calls.
What it actually is:
The interviewer can't show their face because they don't exist. Either it's a text-to-speech operation, a deepfake that can't handle real-time interaction, or the 'recruiter' is running multiple scams simultaneously.
"Fast-Track Hiring"
What it sounds like:
The company urgently needs talent. They're streamlining the process to get you onboarded quickly.
What it actually is:
Urgency prevents verification. Real companies have multi-stage processes. Scammers need you to run that code before you start asking questions like 'can I verify this on your official website?'
"Starter Project"
What it sounds like:
A small onboarding task to familiarize you with the codebase. Normal for new hires.
What it actually is:
The payload delivery mechanism. The 'project' contains obfuscated code in dependencies that executes on install. You're not fixing a bug — you're installing a backdoor.
"Sign-on Bonus"
What it sounds like:
A welcome payment to sweeten the offer. Common in competitive tech hiring.
What it actually is:
The carrot that makes you ignore red flags. You'll never see it. But the promise of $10K-$20K upfront makes you less likely to question why the hiring process feels rushed.
Every term sounds like standard tech hiring. None of them are real. That's the entire trick.
How They Find You
The scam doesn't start with 'clone this repo.' It starts with flattery.
LinkedIn Outreach
A recruiter from a recognizable company (or a convincing clone) messages you. They've 'seen your GitHub' and are impressed. The role matches your skills perfectly. The salary is 20-40% above market.
They researched you. That's what makes it personal. That's what makes it work.
Telegram/Discord Contact
Someone saw your contributions to a Web3 project. Their client is hiring. They can fast-track you. The conversation moves to private channels where there's no audit trail.
Off-platform means off-record.
Job Board Posting
A legitimate-looking job posting for a remote Web3 developer role. Great salary. Flexible hours. Quick process. You apply. They respond immediately.
Real jobs have competition. Fake jobs have immediate availability.
Real Scam Messages
These are actual messages from this operation. The professionalism is manufactured. The opportunity doesn't exist. If you receive anything similar, do not respond:
"Hi! I'm a Senior Recruiter at [Major DeFi Protocol]. We've been reviewing candidates with strong TypeScript/React backgrounds, and your GitHub contributions caught our attention. We're hiring for a Senior Developer role, $180k-$250k base + equity. Would you be open to a quick chat this week?"
Translation: I scraped your LinkedIn. This salary will make you ignore the red flags.
Notice how specific it sounds — TypeScript, React, your GitHub. This isn't a generic message. They research targets to make outreach feel legitimate. The salary range is designed to trigger excitement, not skepticism.
"Sorry, my camera isn't working today — IT issue on my end. Can we proceed audio-only? I'll share my screen for the technical discussion. Really sorry about this, it's been an ongoing problem."
Translation: I can't show my face because I don't exist. Or I'm a deepfake that can't handle questions.
The apology makes it feel human. Real recruiters sometimes have camera issues. But combined with other red flags — no verifiable company presence, rushed timeline — this is a tell.
"Great interview! For the final step, please clone this repo and fix the bug in the payment module. We need it done by tomorrow for our sprint planning. It should take 1-2 hours max. Once you submit the PR, we'll extend the official offer."
Translation: Run npm install on this malware. The deadline prevents you from reading the code.
"1-2 hours" and "by tomorrow" are not about evaluating your skills — they're about preventing due diligence. You won't inspect package.json. You won't check postinstall scripts. You'll run npm install because you want the job.
"Quick update — the hiring manager loved your background but we have another strong candidate in final rounds. If you can complete the assessment today, we can prioritize your offer. Don't want you to miss this opportunity."
Translation: There is no other candidate. There is no offer. Only urgency.
Manufactured competition. The fear of losing a $200K job makes you act fast. Acting fast means not verifying. Not verifying means you run the code.
"Hey, just following up — did you get a chance to clone the repo? Some candidates had issues with dependencies so make sure you run npm install before diving into the code. Let me know if you need help with the setup!"
Translation: Please trigger the malware. The dependencies ARE the attack.
Helpful, friendly, normal. Except the 'dependency issues' they're warning about are the malware working as intended. The follow-up is to ensure you actually execute the payload.
Red Flags
- •Camera always off or 'broken' — In 2025, camera issues are rare. Consistent camera problems across multiple calls is a pattern, not bad luck.
- •Face doesn't track naturally — Deepfakes struggle with head movement, eye contact, and lighting changes. If something feels 'off,' trust your instincts.
- •Audio-video sync issues — Real-time deepfakes have processing delays. If lips don't match audio, you're not talking to a real person.
- •Requests to run unknown code — No legitimate company asks you to run code on your machine before you've verified their identity.
- •Unusually fast hiring process — Real companies have compliance, background checks, and multi-stage interviews. 'We can skip the usual process' means there is no process.
- •Can't verify company on official channels — If you can't find the recruiter on the company's official website or LinkedIn verified employees, they don't work there.
- •Pressure to complete quickly — 'By tomorrow' and 'other candidates interested' are pressure tactics. Real opportunities wait for due diligence.
One red flag is a concern. Three is a scam. They brought the entire set.
The Math That Should Stop You
Let's run the numbers on why this deal makes no sense:
The Salary Problem
They're offering $180K-$250K for a role they're filling through cold LinkedIn messages. Real companies at this salary level use established recruiting firms, referrals, and multi-stage processes. They don't DM strangers.
The Speed Problem
Interview to offer in 48 hours for a six-figure role. Real hiring at this level involves HR, legal, background checks, reference calls, and equity discussions. None of which can happen in two days.
The Verification Problem
If the company is real, why can't you find this role on their careers page? Why isn't the recruiter verifiable? Why does all communication happen off-platform? Because it's not real.
How This Scam Actually Works
The scam follows a precise pattern. Each step sounds like normal tech hiring. None of them are.
Step 1: The Outreach
Recruiter contacts you via LinkedIn, Telegram, or Discord with a role that matches your profile perfectly. High salary. Remote friendly. Interesting project.
They researched you. It feels personal because it's designed to.
Step 2: The Interview
Video call happens. Camera is off or the face seems slightly artificial. The conversation is professional. They ask real technical questions. You feel evaluated.
2025 deepfakes still have tells. But you're too focused on the opportunity to notice.
Step 3: The Assessment
You're asked to clone a GitHub repo and 'fix a bug' or 'build a small feature.' There's a deadline. The faster you complete it, the faster you get the offer.
The deadline prevents code review. That's the point.
Step 4: The Execution
You run npm install. Hidden postinstall scripts execute. Your browser extension data — including wallet credentials — is silently extracted and sent to attacker servers.
The malware runs once. That's all it needs.
Step 5: The Drain
Within hours, your crypto assets are transferred to attacker wallets. The 'recruiter' stops responding. The GitHub repo is deleted. The opportunity never existed.
They'll contact someone else by tomorrow.
Real Cases
These aren't hypotheticals. These are documented victims.
Three developers. Three skill levels. Three reasons for wanting a better opportunity. One outcome: wallets drained, jobs that never existed. The scam doesn't target the naive. It targets the qualified.
The Senior Developer ($85,000)
A 10-year veteran ran a 'code assessment' that looked like a standard React project. The postinstall script extracted his MetaMask seed phrase. Wallet drained in 4 hours.
"I've reviewed hundreds of repos. I never check package.json postinstall scripts. Now I always will."
The CTO ($320,000)
Interviewed for a 'VP of Engineering' role at a DeFi protocol. Demo code for the 'final presentation' contained a hidden payload. His hardware wallet wasn't connected — that saved the rest of his portfolio.
"They knew exactly who I was. The role was designed for my background. That's what made it convincing."
The Designer ($45,000)
Received a 'UI test' that required installing npm packages to view the design system. The packages contained a browser extension scanner that found and exfiltrated wallet data.
"I'm not even a developer. I just wanted to preview some components. I didn't know npm could do that."
Profiles are composites based on documented attack patterns, not individual victims.
Why Developers Fall For It
The victims aren't careless. They're experienced. There's a difference.
The Perfect Role
When a job matches your skills exactly and offers 40% more than market rate, excitement overrides skepticism. You want it to be real.
Routine Familiarity
Developers run npm install hundreds of times. It's muscle memory. Code assessments are normal. The scam uses normal processes to hide abnormal intent.
Deadline Pressure
'Complete by tomorrow' isn't about speed — it's about preventing verification. When you're rushing, you don't inspect. When you don't inspect, you execute malware.
Invested Emotion
After multiple interviews and time invested, you're emotionally committed. Completing the assessment is easier than admitting you might have been played.
This doesn't require naivety. Just hope. Just timing. Just one 'npm install' without reading package.json.
How to Protect Yourself
There are five rules. They will save you every time.
- Rule 1: Verify the company independently Email the company directly using contact info from their official website. Never trust contact info the recruiter provides. If the recruiter can't be verified on LinkedIn by other confirmed employees, they don't exist.
- Rule 2: Never run code on your main machine Use a VM or sandbox for any technical tests. Assume all test code is malicious until you've reviewed it line by line. Your development machine should never have wallet extensions with real assets.
- Rule 3: Demand camera-on interviews Legitimate recruiters show their faces. Consistent camera issues across multiple calls is a red flag. If they refuse, end the conversation.
- Rule 4: Read package.json before npm install Check postinstall scripts. Check dependencies for obfuscated names. If anything looks unusual, don't run it. Real companies don't hide code in dependencies.
- Rule 5: Question urgency Real opportunities don't disappear if you take a day to verify. 'Other candidates' and 'tomorrow's deadline' are manipulation tactics. Legitimate companies respect due diligence.
Think You Can Spot a Scam?
Test your knowledge with real scam scenarios from our investigations.
The Technical Kill Chain
Understanding exactly how the malware executes helps you spot it before it's too late:
Stage 1: Obfuscated Dependencies
The package.json references a dependency with an innocuous name like 'react-utils-helper' or 'webpack-dev-config'. These packages exist on npm but are controlled by attackers. Version pinning looks normal: '^2.1.4'.
Stage 2: Postinstall Execution
The malicious package contains a 'postinstall' script that runs automatically when npm install completes. This script is often base64-encoded or uses character code obfuscation to avoid detection.
Stage 3: Browser Data Extraction
The payload targets browser extension directories for MetaMask, Phantom, and other wallets. It extracts LevelDB files containing encrypted vault data, session tokens, and browser cookies.
Stage 4: Credential Exfiltration
Data is sent to attacker-controlled servers via HTTPS POST requests to hostnames chosen to look like ordinary infrastructure — generic analytics or CDN subdomains. We name none of them: any specific hostname we could print here would be an invented illustration, not a confirmed indicator. The request completes in milliseconds — you won't notice network activity.
Stage 5: Key Derivation Attack
Attackers use the extracted vault data combined with common password lists to crack wallet encryption. Weak passwords fall in minutes. Once they have your seed phrase, your wallet is theirs forever.
npm install runs. Wallet data extracted. 15 seconds. You're still reading the README.
Inside the Attacker's Playbook
These aren't amateur operations. They're organized criminal enterprises with professional infrastructure:
- •Fake Company Websites — Professional landing pages, team bios with AI-generated headshots, job listings that match real market demand.
- •LinkedIn Automation — Profiles aged 2+ years, real-looking connection networks, endorsements from other fake accounts.
- •GitHub Presence — Repos with commit history, forks of popular projects, contributions that look legitimate.
- •Domain Infrastructure — Domains registered months in advance, SSL certificates, email servers that pass SPF/DKIM checks.
- •Money Laundering Pipeline — Extracted crypto immediately mixed through Tornado Cash, cross-chain bridges, and converted to Monero.
They invested more in convincing you than most startups invest in product.
2026 Update: LinkedIn Executive Recruiting Scams
The scam has evolved beyond developer roles. In early 2026, attackers are targeting executives and compliance professionals with sophisticated LinkedIn messages for senior positions at established foundations and protocols.
- ▸Role titles now target non-technical professionals: 'Global Head of Tokenomics,' 'Regulatory Strategy Lead,' 'Chief Compliance Officer.' The salary ranges ($150K–$240K+) are designed to attract senior talent.
- ▸Messages are AI-generated but hyper-personalized — referencing the target's certifications, advisory mandates, and geographic context (e.g., EEA regulatory navigation). The flattery is surgical.
- ▸Red flags remain consistent: salary advertised upfront in the initial message, zero mutual connections with the recruiter, messages arriving outside normal business hours, and profiles with minimal activity or engagement from unrelated geographies.
- ▸The attack chain has expanded: after engagement, targets are asked to 'send your CV' (data harvesting for future attacks), complete a 'quick interview' on an external platform (credential phishing via fake SSO), or install a 'video client' (malware delivery).
- ▸End goals include: phishing credentials via fake SSO login pages, pushing malicious browser extensions disguised as video clients, hijacking calendar and email flows through OAuth permissions, and ultimately extracting 'verification fees,' 'background check payments,' or 'equipment deposits.'
Got a Suspicious Message?
Use our AI-powered detector to analyze potential scams instantly.
Keep digging
If you have the message, link or contract in front of you right now, run it through the free AI scam detector, which scores the text, the lookalike domain and the contract and hands back a permanent report you can share.
Machine-speed versions of these scripts are catalogued in our hub on AI-powered crypto scams.
The customer list is the next phishing campaign — that chain is documented in our hub on data breaches and vendor leaks.
Key Takeaways
- 1.Camera-off interviews in 2025 are a major red flag — technology rarely fails this consistently.
- 2.Never run unknown code on a machine with crypto wallet access. Use VMs. Always.
- 3.Verify recruiters through official company channels, not through info they provide.
- 4.Legitimate companies don't rush technical assessments. Urgency is a manipulation tactic.
- 5.The malware isn't in the code you review — it's in the postinstall scripts you never check.
“The interview was real.”
The job was the scam.
Frequently Asked Questions
Related case: Fake Binance support calls and impersonation.
Sources & Citations
Research for this investigation compiled from publicly available blockchain data, security reports, and community documentation.
www.recordedfuture.com · Date not published or not verified
Chainalysis · 17 Jul 2025
Verification: On-chain references: none published for this case.
Methodology: Every case needs at least two independent sources before publication, plus verifiable on-chain evidence whenever a public transaction trail exists. Full standards: /methodology
Legal notice: This assessment is based on publicly available data, including on-chain records, official statements and reported incidents. It is journalistic and educational analysis, not legal advice, an accusation of criminal conduct or a court finding. Named companies, projects, domains, wallets and individuals are described as reported by the cited sources; a company name may appear because fraudsters impersonated it, not because the company did anything wrong. If you believe something is inaccurate or out of date, write to cryptostrapon@pm.me and we will correct it and log the change. Editorial policy