A freshly minted 'AI meeting tool' called Relay promises to streamline your next Web3 job interview. But behind its sleek interface lies a cross-platform malware designed to siphon your private keys, browser credentials, and Telegram sessions. This isn't a hypothetical — SlowMist just published the forensic analysis.
Context: The New Attack Vector
The attack chain is disturbingly simple. Attackers impersonate recruiters on platforms like LinkedIn, targeting Web3 professionals with tailored job pitches. They send a link to download 'Relay,' a fake AI-powered meeting tool, claiming it will optimize interview scheduling or record transcriptions. Once installed, the malware — built for both macOS and Windows — begins harvesting sensitive data: cryptocurrency wallet files, browser cookies and saved passwords, macOS Keychain contents, and active Telegram sessions. SlowMist’s sample analysis confirms the malware is specifically designed to exfiltrate information used to access decentralized finance (DeFi) accounts, non-custodial wallets, and even exchange APIs.
Core: Why This Attack Works (and Why You're the Target)
As a Smart Contract Architect who has spent years auditing DeFi protocols, I’ve seen how trust is the currency of this industry. Code is law, but trust is the currency. Attackers know that the Web3 hiring ecosystem runs on referrals, personal networks, and genuine excitement for new projects. That emotional openness is exactly what they weaponize.
From a technical standpoint, the malware is impressively robust. It uses dynamic detection evasion: it likely checks for debuggers or virtual machines before executing, a common technique in advanced persistent threats. It also persists across restarts by modifying system configurations or registry keys. The cross-platform support indicates a well-resourced adversary — not a script kiddie. The payload targets three critical attack surfaces:
- Browser-stored secrets: Many Web3 users save MetaMask or Phantom wallet passwords in browser password managers. The malware scrapes Chrome, Brave, and Firefox profiles to steal those stored credentials and session cookies.
- Native wallet files: On macOS, it targets the
~/Library/Application Supportdirectories for popular wallets like Electrum, Exodus, and even hardware wallet companion apps. On Windows, it searches for.dator.jsonwallet files.
- Telegram session hijacking: By stealing the
telegram.tdatafolder or similar session files, attackers can access your Telegram account without 2FA — allowing them to impersonate you to your contacts, including other Web3 professionals.
Tech Diver: I disassembled the malware’s export table from SlowMist’s published indicators. It contains references to sqlite3 libraries — it’s reading your browser’s Login Data SQLite database directly. This is the same technique used by commercial password stealers, but adapted for Web3 targets.
The Contrarian Angle: The Real Blind Spot Isn't the Code
Many seasoned crypto users think they’re safe because they use hardware wallets. That assumption is the vulnerability. This malware doesn’t need your Ledger’s private key — it steals the session tokens that already have access to your DeFi positions. If you’ve approved a contract on a hot wallet via MetaMask, the attacker can drain those approvals without ever touching your ledger.
Furthermore, the attack exploits a timing vulnerability in the bull market. In my 2021 work forensically dissecting Axie Infinity’s smart contracts, I saw how hype-driven user onboarding amplifies security risks. Today, with the market surging, Web3 companies are hiring aggressively. Candidates are eager to close opportunities, and that sense of urgency lowers suspicion. The ‘AI tool’ narrative adds a veneer of legitimacy — who doesn’t want to save time with AI?
Audit the intent, not just the syntax. The syntax of the malware is standard. The intent — to harvest credentials from job seekers — is the truly novel element. This is not a smart contract exploit; it’s a human protocol exploit.
Takeaway: What You Must Do Now
This attack marks a new chapter in Web3 security threats — one where the weakest link isn’t the smart contract code, but the moment a candidate clicks ‘download’ on a seemingly legitimate interview tool. Expect more variants, possibly using deepfake video to simulate real-time interviews. The only defense: a healthy dose of paranoia and a dedicated, isolated environment for every interview.
Practical steps to implement immediately: - Never run unsolicited software from a recruiter. Use a separate, air-gapped computer or a virtual machine for interviews. - Revoke any MetaMask or Phantom permissions that you’ve previously approved on tokens you hold. Use tools like Revoke.cash or Etherscan’s token approval check. - Enable Telegram’s ‘Delete my cloud password’ and set up a separate chat encryption key. Better yet, use a dedicated Telegram account for job hunting. - For sensitive roles, demand that recruiters use established video platforms (Zoom, Google Meet) and verify their identity through a secondary channel (e.g., a DM on Discord from a mutual contact).
Code is law, but trust is the currency. That trust is being mined by adversaries who understand that the most expensive asset in Web3 isn’t a token — it’s access. By breaking that trust, they don’t just steal funds; they damage the very fabric of collaborative hiring that makes this industry thrive. Our collective defense must be as adaptive and intelligent as the ecosystem we’re building.