A bug bounty program is a structured system where:
Rather than publicly exposing a vulnerability, researchers report it privately to the organization first, giving them time to fix it.
| Feature | Private Program | Public Program |
|---|---|---|
| Who can participate | Invited researchers only | Anyone |
| Report volume | Lower, controlled | Higher, broader |
| Noise level | Low | Higher |
| Discovery potential | Limited | Maximum |
2022 → Private program (HackerOne VIP)
→ Thousands of vetted researchers
→ Refined processes
Now → Single unified PUBLIC program
Vercel's reasoning teaches us the tradeoffs organizations consider:
They built custom tooling to filter noise and streamline triage before going public — an important lesson: scale your processes before scaling access
This is a new and important dynamic to understand:
AI Impact on Bug Bounty Programs
├── More researchers can find vulnerabilities (lower skill barrier)
├── More reports submitted overall
├── Both VALID and INVALID reports increase
└── Organizations must choose:
├── Go private (reduce noise) ← Some companies
└── Go public + build better filters ← Vercel's choice
Key Insight: AI democratizes security research — both a challenge and an opportunity.
HackerOne is a third-party platform that:
Researcher finds vulnerability
↓
Submit report on HackerOne
(with clear reproduction steps)
↓
Security team triages report
↓
Valid? → Fix developed & shipped
↓
Researcher rewarded + fix verified
Scope defines the boundaries of what researchers are allowed to test.
| Concept | Key Takeaway |
|---|---|
| Bug Bounty | Paid program for responsible vulnerability disclosure |
| Private vs Public | Tradeoff between control and coverage |
| Scope | Defines legal and ethical testing boundaries |
| HackerOne | Third-party platform managing the process |
| AI's Impact | Increases report volume — requires better filtering |
| Responsible Disclosure | Report privately, allow time to fix |
Ask yourself: