Video summary
What is Access Control and How Can You Break It?
Main summary
Key takeaways
Summary
Broken access control is a useful vulnerability class for beginner bug bounty hunters. It generally requires little specialized tooling or deep technical knowledge, though finding flaws can take patience and manual testing. The central question is whether a user can access or change something they should not be allowed to.
Access-control models and common flaws
- Role-based access control (RBAC): Permissions depend on a user’s role, such as moderator or editor. A flaw might let an ordinary user perform an admin-only action.
- Discretionary access control: A resource owner chooses who can access something, often through sharing settings. A flaw could expose a resource to someone who was not granted access.
- Attribute-based access control: Permissions are assigned through combinations of roles, users, or other attributes. This model can be flexible but complicated to manage.
Common forms of broken access control include:
- IDOR / BOLA (insecure direct object reference / broken object level authorization): A user changes a resource identifier in a request and accesses another user’s object because the server does not properly check ownership or permissions.
- Broken function-level authorization: A user performs actions reserved for another role, such as using an admin function.
- Missing access control: A resource or function lacks a meaningful authorization check, potentially allowing access even when logged out.
- Cross-tenant access: A user in one organization views, modifies, or otherwise affects another organization’s data in a multi-tenant application.
Identifiers may appear in URLs, request parameters, form data, or API requests. UUIDs make identifiers harder to guess, but they do not fix missing authorization. If an ID can be discovered, the application still needs to verify that the requester has permission to use it.
Testing guide and demonstration
A suggested manual workflow:
- Create test accounts with different roles or resource ownership.
- Use the application and capture requests in Burp Suite.
- Remove the session cookie and check whether requests still work, simulating a logged-out user.
- Replay one account’s request while authenticated as another account. Check whether the action affects the original account’s resource.
- Test admin actions as a regular user and, where relevant, test across organizations.
- Verify the result in the application. A response that looks successful does not necessarily mean the data changed.
Separate browser profiles or Firefox Multi-Account Containers can help keep accounts distinct while routing traffic through Burp. Testing can be time-consuming, and an error or a successful-looking response alone may not establish whether a vulnerability exists.
In the Indeed demonstration, the presenter examines API requests and tries logged-out and cross-account tests. The displayed attempts do not change the other account’s profile, so the demonstration does not establish a vulnerability.
A separate anecdote describes a bug bounty finding on a social media site. Although another user could not edit a supposedly private video, its title, thumbnail, description, and tags were still exposed. The presenter says the issue earned a $500 bounty, illustrating that unauthorized viewing can matter even when unauthorized editing is blocked.
Other information
- Bugcrowd is identified as the sponsor of the course series. The presenter mentions a newly launched Bugcrowd program with OpenAI and encourages viewers interested in AI to consider it.
- The homework is to create two test accounts on Indeed and manually check whether account A’s resources can be accessed from account B.
Main speakers and sources
- Speaker: The unnamed presenter and bug bounty instructor.
- Sources and examples: Bugcrowd (sponsor and platform), Indeed (testing demonstration), Burp Suite, and Firefox Multi-Account Containers.
Rate this summary
Your feedback will help improve summaries.
Improve this summary
Reprocess with a stronger model when the summary feels incomplete or inaccurate.
Translate summary in another language
Ask questions to this video
Chat for follow-up questions, clarifications, and source-backed answers.