The OWASP API Security Top 10
API Security Fundamentals · API Security track
6 min read
In this module
A Framework for API Risks
The OWASP API Security Top 10 is the most-cited classification of API security risk on the planet. First published in 2019 and refreshed in 2023, it gives everyone a shared vocabulary for talking about the most serious API threats. The list is worth knowing because it shapes how organisations prioritise their defences and how security testers structure their assessments.
The 2023 edition tracks how the API threat landscape has shifted since 2019: three categories added, three renamed, two dropped, and one merged from two separate entries in the original list.
The 2023 API Security Top 10
API1: Broken Object Level Authorisation (BOLA)
OWASP ranks BOLA first, and it's widely reported as the most common API flaw; vendor telemetry often puts it behind something like 40% of observed API attacks, though such figures depend on whose traffic is being counted. It happens when an endpoint accepts an object identifier from the client (a user ID, an order number, etc.) without checking that the requesting user is authorised to touch that specific object.
An attacker just swaps the ID in the request and reads someone else's data. The next module digs into BOLA properly.
API2: Broken Authentication
A grab-bag of weaknesses in the authentication mechanism itself: weak password policies, missing brute-force protection, credentials shoved into URLs, tokens that never expire, and JWT implementation bugs like the none-algorithm attack or algorithm confusion.
API3: Broken Object Property Level Authorisation
A merged category from the 2019 list, fusing what used to be "Excessive Data Exposure" and "Mass Assignment". Two sides of the same coin: the API returns properties the user shouldn't see (excessive data exposure), or accepts properties the user shouldn't be allowed to set (mass assignment).
{ "user": { "name": "Alice", "email": "alice@example.com", "role": "admin", "internal_id": "a8f3c2e1", "salary": 95000 }}In the example above, role, internal_id, and salary look like fields the API has no business exposing to every caller.
API4: Unrestricted Resource Consumption
Renamed from "Lack of Resources & Rate Limiting" in 2019, this broader category covers any failure to cap what an API consumer can chew through: request rate, payload size, records returned, file upload size, compute-heavy operations.
API5: Broken Function Level Authorisation (BFLA)
Where BOLA is about access to data objects, BFLA is about access to functions. An attacker reaches administrative endpoints, invokes functions meant for a different role, or fiddles with HTTP methods to slip past access controls.
DELETE /api/v1/users/1842 HTTP/1.1A regular user has no business calling an admin delete operation, but if function-level authorisation is missing, that request will be honoured.
API6: Unrestricted Access to Sensitive Business Flows
A new category in the 2023 list, focused on APIs that expose business-critical functionality without enough friction to deter automated abuse. Think ticket-scalping bots, coupon-abuse rings, mass account creation. The flaw isn't technical so much as a failure of imagination — not anticipating that legitimate business logic could be weaponised at scale.
API7: Server-Side Request Forgery (SSRF)
Another new addition in 2023. SSRF through APIs shows up when a server-side component fetches a URL the client supplied without validating it. From there, attackers scan internal networks, hit cloud metadata services, or reach internal APIs that were never meant to be externally accessible.
API8: Security Misconfiguration
The catch-all bucket of deployment and configuration sins: missing security headers, error messages that proudly print full stack traces, HTTP methods left switched on for no reason, overly permissive CORS policies, default credentials still in place on API management consoles.
API9: Improper Inventory Management
Renamed from "Improper Assets Management" in 2019, this category tackles the deceptively simple question: do you actually know what APIs you have, and which of them are exposed? Shadow APIs, deprecated endpoints still answering requests in production, and dev or staging APIs that are quietly reachable from the open internet — all welcome here.
Shadow APIs are endpoints that exist in production but are not documented, monitored, or included in security assessments. They often arise from legacy code, abandoned features, or development shortcuts. They represent some of the most dangerous attack surface because no one is watching them.
API10: Unsafe Consumption of APIs
The third new category in 2023. This covers what happens when your API consumes data from third-party APIs without validating it. Trust an upstream service blindly and a compromised or malicious upstream can inject payloads, feed you bad data, or trigger behaviour you never intended.
What Changed from 2019 to 2023
The shifts between the two editions track what attackers have actually been doing:
- Merged: Excessive Data Exposure and Mass Assignment fused into API3 (Broken Object Property Level Authorisation), since they're both fundamentally about property-level access control.
- Renamed: "Broken User Authentication" became API2 (Broken Authentication), covering service and machine identities too, not just users. "Lack of Resources & Rate Limiting" became API4 (Unrestricted Resource Consumption), broadening beyond just rate limiting. "Improper Assets Management" became API9 (Improper Inventory Management).
- New entries: API6 (Unrestricted Access to Sensitive Business Flows), API7 (SSRF), and API10 (Unsafe Consumption of APIs) all appeared for the first time, reflecting the rise of business logic abuse, SSRF in cloud-native architectures, and supply-chain risks across interconnected API ecosystems.
- Removed: Injection and Insufficient Logging & Monitoring both dropped out as standalone categories. Both are still real risks, just folded into the other entries.
Where to Focus
These categories don't carry equal weight. OWASP puts BOLA (API1) first, and it's the class most often reported in API breaches. Broken Authentication (API2) and Broken Object Property Level Authorisation (API3) round out the top three, and in practice they're where most testers find their best bugs, so they're the place to start.
Security misconfiguration (API8) and improper inventory management (API9) are usually the enablers, setting the stage that makes other vulnerabilities discoverable and exploitable in the first place.
In the next lesson we'll set up the tools you need to actually test for these vulnerabilities in a controlled environment.