Skip to content

The Future of ATO

Presenters detail vulnerability discovery and exploitation in The Future of ATO.

Summary

## Overview

Philip Martin (Coinbase; cryptocurrency exchange/brokerage/custodian) presented Account Takeover (ATO) attack trends and defensive strategies based on Coinbase's operational experience. Coinbase operates globally across 103 countries with 30 million users, making it target for some of world's most targeted, high-effort ATO campaigns. Coinbase self-describes as "world's largest capture-the-flag with money on it"—due to cryptocurrency holdings, platform attracts diverse threat actors from solo attackers to nation states (e.g., North Korea). Distinction between targeted and untargeted ATO critical: targeted attacks involve in-depth victim research, active pivoting around security controls, and systematic compromise of account ecosystem; untargeted attacks rely on bulk credential stuffing from leaked databases, moving on instantly if blocked. Defense efficacy depends on attack type: SMS 2FA and TOTP effective against untargeted bulk attacks; less effective against targeted attacker pivoting; Universal 2nd Factor (U2F) / WebAuthn most effective across all threat types. Research validates Google's findings: hardware security keys provide defense resistant to even highly-motivated targeted attackers.

## Targeted vs. Untargeted ATO Attack Models

**Targeted Account Takeover (10% of observed attacks at Coinbase)**: High-effort, high-reward operations targeting wealthy cryptocurrency holders. Attackers conduct in-depth reconnaissance on victims—researching social media presence, public statements, wealth indicators (e.g., photos with Bitcoin symbols, public crypto discussions). Attack process involves systematic compromise of account ecosystem before accessing target's Coinbase account. Attackers first establish toehold by compromising email account (highest-value target providing password reset access), then pivot to file-sharing services (Dropbox, Box), social media platforms, and other online accounts. By time attacker reaches cryptocurrency holdings, victim's entire online identity compromised. Attackers actively pivot around security controls—encountering roadblock in one attack vector triggers pivot to alternative approach rather than abandonment of target. Distinguishes targeted ATO from commodity attacks: attackers remain committed to specific victim until successful.

**Untargeted Account Takeover (90% of attacks; 50+ billion attempts reported)**: Low-effort, low-reward bulk credential attacks using leaked password databases and scraped credentials. Attackers lack knowledge of target identities or assets. Attackers simply process credentials in bulk, moving instantly to next target upon encountering security controls (no pivoting, no persistence). Effort/reward ratio extremely unfavorable for individual attempts; profitability derives from volume—even 0.1% success rate against billions of attempts generates significant payoff. No reconnaissance or targeting; attackers indifferent to victim identity or cryptocurrency holdings.

**Attack Actor Spectrum**: Coinbase encounters attacks from solo attackers (opportunistic but persistent), organized criminal syndicates (dedicated teams, infrastructure), transnational crime organizations (coordinated across borders), and nation states (North Korea notably active in cryptocurrency theft). Threat sophistication and resources scale across spectrum; nation states employ advanced techniques beyond typical criminal capabilities.

## Defensive Measures and Risk Mitigation

**2FA/MFA Effectiveness Against Attack Types**: SMS 2FA and TOTP (Time-based One-Time Password) highly effective against untargeted bulk attacks. Attacker possessing username/password from leaked database typically lacks access to victim's SMS or authenticator app, preventing account takeover. However, targeted attackers with in-depth access to victim's account ecosystem (including compromised email) can redirect SMS messages (SIM swapping), intercept authentication codes, or bypass TOTP via authenticator app compromise. Effectiveness degrades under targeted attack model.

**Universal 2nd Factor (U2F) / WebAuthn Superiority**: Hardware security keys (U2F/WebAuthn) resistant to both untargeted and targeted ATO attacks. Unlike SMS/TOTP (which can be intercepted/redirected), U2F operates via cryptographic proof-of-possession requiring physical possession of registered key. Phishing attacks (pretending to be legitimate service) fail against U2F as key validates server certificate and won't release authentication credential to wrong domain. Targeted attackers cannot bypass U2F without physical theft of security key. Research findings (Coinbase and Google) converge: U2F provides exceptional defense across threat spectrum.

**Risk Stratification for Cryptocurrency Users**: Public activity identifying holders as cryptocurrency-wealthy creates targeted attack vulnerability. Risk factors: posting social media about crypto holdings, photographing oneself with Bitcoin symbols, engaging in public cryptocurrency discourse, creating publicly-traceable crypto activity trail. Low-risk users (private about holdings, no public footprint) effectively protected by SMS 2FA against typical bulk attacks. High-risk users (public crypto advocates, wealthy identifiable holders) should implement U2F to defend against targeted attackers researching and compromising them specifically. Security posture should match threat model.

**Messaging and User Education**: Coinbase's diverse global user base (age, experience, geography) complicates security communication. Users unfamiliar with cryptography struggle to understand U2F's superiority over SMS; perceived usability burden vs. perceived (but illusory) convenience of SMS 2FA affects adoption. Education gap remains challenge: convincing users of targeted attack risk when they perceive themselves as low-value targets.

Related talks

Black Hat·4 min

  • Black Hat
  • Community
  • Date: 2025
  • Location: Black Hat

Engineers explore architecture and security mechanisms for Welcome Video.

welcomevulnerability-researchthreat-detection

Black Hat·26 min

  • Black Hat
  • Community
  • Date: 2025
  • Location: Black Hat

Electromagnetic Fault Injection · Automotive Systems

Electromagnetic Fault Injection examines Bam the BAM.

vulnerability-researchthreat-detectiondefensive-security

Black Hat·49 min

  • Black Hat
  • Community
  • Date: 2025
  • Location: Black Hat

Poisoned RDP Offense

Poisoned RDP Offense investigates He Said, She Said.

vulnerability-researchsaidthreat-detection