AI summary by TwinMind
The Moonpig Bug: How 3,000,000 Customers' Details Were Exposed
Tom Scott
5m 26s
Security Vulnerability Discovery
- Developer Paul Price disclosed major security hole in Moonpig (British personalized products company)
- Vulnerability exposed all users' personal details sufficient for identity theft
- Paul responsibly disclosed privately, gave 1 year warning before publishing (11 months more than needed)
- Only tested on himself and test accounts he created
How Proper App Security Works
- User logs in with username/password over secure connection
- Server provides random token (like browser cookie) for future authentication
- Token replaces password storage on device
- Third-party apps use OAuth2 for additional security layers
- Tokens should be revocable when passwords change
Moonpig's Fatal Security Flaws
- Used customer ID as authentication token instead of random string
- Customer IDs were sequential numbers (51624, 51623, 51625)
- No rate limiting on API requests
- No monitoring for excessive requests from single source
- Attacker could impersonate any user by guessing their ID
Exposed Data and Attack Potential
- Name on credit card
- Last 4 digits of credit card
- Expiry date
- Full address
- Historical precedent: enough data to compromise iCloud accounts via Apple support
- Could lead to Amazon account takeover and $10,000 bitcoin mining charges
Mass Data Extraction Capability
- Sequential IDs allowed automated scraping of all customer data
- No protection against bulk data harvesting
- Unknown if malicious actors exploited this during 1-year disclosure period
Company Response and Lessons
- Moonpig PR statement: "all password and payment information is and always has been safe"
- Suspiciously specific denial - didn't address addresses, impersonation, or other claims
- No response to follow-up questions about last 4 digits classification
- Shut down app API after disclosure
- 7+ hours no reply to detailed questions
Developer Security Recommendations
- Code defensively assuming active attacks
- Never trust user input
- Assume malicious intent until proven otherwise
- Think "how could I break this?" during development
- Use proper random tokens, not sequential IDs
- Implement rate limiting and request monitoring
Non-Technical Management Advice
- Hire external security testing (search term: "pentest")
- When security bugs reported: fix immediately, say thank you
- Don't ignore or spin security issues
- Budget for outside security audits
