Skip to main content
Be Bitwise

Scope, Rules & Legal Boundaries

Bug Bounty Fundamentals · Bug Bounty track

6 min read

In this module

The Rules That Protect Everyone

Bug bounty hunting means testing systems you don't own. Strip away the rules and legal protections and the activity is indistinguishable from criminal hacking. Scope definitions, rules of engagement, and safe harbour clauses are how the researcher and the organisation both stay out of court. Understanding these boundaries is the price of entry.

In-Scope vs Out-of-Scope Assets

Every programme defines exactly which assets you are authorised to test. In-scope assets are explicitly listed and may include:

  • Specific domains and subdomains (e.g., *.example.com)
  • Mobile applications (iOS, Android)
  • API endpoints
  • Open-source repositories
  • Specific IP ranges

Out-of-scope assets are explicitly excluded. Common exclusions include:

  • Third-party services (e.g., a payment processor's domain)
  • Corporate infrastructure (email servers, VPN endpoints)
  • Customer data and production databases
  • Specific subdomains used for internal operations
  • Partner or vendor systems
text
# Example scope definition
In Scope:
*.example.com
api.example.com
Example iOS App (App Store)
Example Android App (Google Play)
Out of Scope:
mail.example.com
vpn.example.com
status.example.com
Any third-party services (Stripe, AWS, etc.)

If an asset isn't explicitly listed as in-scope, treat it as out of scope. When in doubt, ask the programme team before you test.

Rules of Engagement

Beyond the asset list, programmes define which methods you can use. These are the rules of engagement, and they apply no matter what you find:

Almost always prohibited:

  • Denial of service (DoS/DDoS) testing
  • Social engineering, phishing, or pretexting against employees
  • Physical access testing
  • Accessing, modifying, or exfiltrating real user data
  • Automated scanning that generates excessive traffic
  • Testing against other users' accounts without explicit permission

Often restricted:

  • Automated vulnerability scanning (some programmes allow it; many do not)
  • Chaining attacks that modify production data
  • Accessing internal systems beyond what is necessary to demonstrate the vulnerability
The Minimum Necessary Principle

When demonstrating a vulnerability, use the minimum level of access needed to prove the issue exists. If you find an IDOR that exposes user data, access one record to confirm the flaw - do not dump the entire database. If you discover a remote code execution vulnerability, run a harmless command like id or whoami - do not install persistence or pivot to other systems. The goal is to prove the vulnerability, not to exploit it fully.

Safe Harbour Clauses

A safe harbour clause is the organisation's written authorisation for the testing it describes, plus a promise that it won't take legal action against you or report you to the authorities, provided you played by the programme rules. It's one of the most valuable things a programme can offer.

Its reach has limits. A company can't bind the police or the CPS, so it can't promise you'll never be prosecuted, only that it authorised you and won't complain. And it can authorise testing only of systems it controls: a safe harbour from the programme doesn't cover a third-party host, cloud provider or partner service that sits behind the target.

A strong safe harbour clause typically includes:

  • A commitment not to initiate legal action for authorised testing
  • A promise to work with researchers in good faith
  • A statement that the programme constitutes authorisation under applicable computer crime laws
  • Guidance on what constitutes "good faith" testing

Not every programme offers one. Some give you only a vague statement of non-prosecution; others give you nothing at all. Read the policy carefully before you test. Programmes hosted on major platforms like HackerOne generally include solid safe harbour language by default.

In the United Kingdom, the primary legislation governing computer security research is the Computer Misuse Act 1990 (CMA). The CMA creates five offences (covered in detail in the module on responsible disclosure and the law). The three that matter most for bug bounty work are:

  1. Unauthorised access to computer material (Section 1)
  2. Unauthorised access with intent to commit further offences (Section 2)
  3. Unauthorised acts with intent to impair, or with recklessness as to impairing, the operation of a computer (Section 3)

The other two are Section 3ZA (unauthorised acts causing, or creating a risk of, serious damage) and Section 3A (making, supplying or obtaining articles for use in CMA offences).

Everything turns on "unauthorised." A programme's scope and safe harbour clause serve as explicit authorisation for the testing they describe. Stay inside that envelope and you're acting lawfully; step outside scope or use prohibited methods and you may have just lost that protection.

Other jurisdictions have their own equivalents. The US has the Computer Fraud and Abuse Act (CFAA), the EU has the Directive on Attacks Against Information Systems, and most countries have something on the books. Know the law where you live, and the law where the target lives.

What to Do If You Accidentally Go Out of Scope

It happens. You follow a redirect into a third-party domain, find that a subdomain resolves to a shared host, or realise too late that the API endpoint you're hitting belongs to a partner service.

If this happens:

  1. Stop testing immediately on the out-of-scope asset
  2. Document what happened - record the exact steps that led you there
  3. Do not exploit or investigate further on the out-of-scope system
  4. Notify the programme through the platform, explaining what occurred
  5. Delete any data you may have inadvertently accessed

Honest, prompt disclosure of an accidental scope slip is almost always handled professionally. Trying to hide it, or carrying on testing the out-of-scope asset, is how an honest mistake metastasises into a legal problem.

Programmes remember the researchers who own up to their mistakes. A quick, transparent self-report buys you far more trust than silently pushing the envelope.