The OWASP Top 10 is a standard awareness document from the Open Worldwide Application Security Project (OWASP) listing what the project calls, in its own words, "a broad consensus about the most critical security risks to web applications." OWASP Top 10:2025 is the current, eighth edition — following 2003, 2004, 2007, 2010, 2013, 2017 and 2021 — announced as a release candidate at the OWASP Global AppSec Conference in November 2025, with OWASP's own GitHub repository README now describing it as released and "Final."
Key takeaways
- A01:2025 Broken Access Control holds the #1 spot it also held in 2021.
- New: A10 Mishandling of Exceptional Conditions, with 24 CWEs on "improper error handling, logical errors, failing open," partly drawn from CWEs OWASP previously saw as poor code quality.
- OWASP's own "What Are Application Security Risks" page describes a hybrid process.
- Each category page opens with the same nine figures.
What are the 10 categories?
Per OWASP's own published list, with the CWE count and average incidence rate from each category's score table on GitHub:
| Code | Category | 2021 rank | CWEs mapped | Avg incidence |
|---|---|---|---|---|
| A01:2025 | Broken Access Control | #1 (now includes 2021's A10 SSRF) | 40 | 3.74% |
| A02:2025 | Security Misconfiguration | #5 | 16 | 3.00% |
| A03:2025 | Software Supply Chain Failures | new (expanded from A06:2021) | 6 | 5.72% |
| A04:2025 | Cryptographic Failures | #2 | 32 | 3.80% |
| A05:2025 | Injection | #3 | 37 | 3.08% |
| A06:2025 | Insecure Design | #4 | 39 | 1.86% |
| A07:2025 | Authentication Failures | #7 (renamed from "Identification and Authentication Failures") | 36 | 2.92% |
| A08:2025 | Software or Data Integrity Failures | #8 (renamed from "Software and Data Integrity Failures") | 14 | 2.75% |
| A09:2025 | Security Logging & Alerting Failures | #9 (renamed from "Security Logging and Monitoring Failures") | 5 | 3.91% |
| A10:2025 | Mishandling of Exceptional Conditions | new | 24 | 2.95% |
A01:2025 Broken Access Control holds the #1 spot it also held in 2021. OWASP's own category page describes it as failures where "users cannot act outside of their intended permissions," covering violations of least privilege, URL-parameter tampering, insecure direct object references, missing API-level authorization checks, JWT/cookie tampering, CORS misconfiguration and forced browsing. It is now the single largest category by scope: OWASP's own data states 40 CWEs map to it, with a maximum incidence rate of 20.15% across tested applications, 1,839,701 total occurrences and 32,654 total CVEs — and 2021's standalone Server-Side Request Forgery (SSRF) category was folded into it for 2025.
A03:2025 Software Supply Chain Failures is the edition's most consequential new category. OWASP's own text describes it as "breakdowns or other compromises in the process of building, distributing, or updating software," evolved from 2021's narrower "Vulnerable and Outdated Components" (itself tracing back to 2013's "Using Components with Known Vulnerabilities"). Its score table lists 6 CWEs (the introduction says 5), level with A09 as the narrowest categories, and OWASP's data notes it carries the highest average exploit and impact scores among mapped CVEs (8.17 and 5.23). It ranked #1 in OWASP's community survey with exactly 50% of respondents placing it first, although only 11 CVEs carry its CWEs. OWASP's own category page cites the 2019 SolarWinds compromise (~18,000 organizations affected through a trusted vendor) and the 2025 npm "Shai-Hulud" worm (self-propagating across 500+ package versions) as example attack scenarios.
The category totals in the score tables add up to 249 CWE mappings; the introduction gives 248, and where a category page and the introduction differ by one, this guide quotes the category page.
What changed from 2021, category by category?
OWASP's introduction states there are "two new categories and one consolidation." In detail:
- Promoted: Security Misconfiguration rose from #5 to #2; OWASP links this to "the amount of an application's behavior that is based on configurations."
- New: A03 Software Supply Chain Failures, an expansion of A06:2021 to "compromises occurring within or across the entire ecosystem of software dependencies, build systems, and distribution infrastructure."
- New: A10 Mishandling of Exceptional Conditions, with 24 CWEs on "improper error handling, logical errors, failing open," partly drawn from CWEs OWASP previously saw as poor code quality.
- Consolidated: SSRF (A10:2021) folded into A01; CWE-918 is now one of A01's notable CWEs.
- Moved down: Cryptographic Failures #2 → #4, Injection #3 → #5, Insecure Design #4 → #6. For Insecure Design, OWASP reports "noticeable improvements in the industry related to threat modeling."
- Renamed: A07, A08 (the A08 page calls "and" → "or" a "slight, clarifying name change") and A09, where "Alerting" replaces "Monitoring" because "great logging with no alerting is of minimal value."
The 2025 edition went final on GitHub in December 2025: commit f7439d4, dated 24 December 2025, removes the release-candidate styling "now that 2025 is going live," and the repository README marks 2021 as "SUPERSEDED."
How were the categories chosen?
OWASP's own "What Are Application Security Risks" page describes a hybrid process. The quantitative side draws on roughly 175,000 CVE records mapped to CWEs from the National Vulnerability Database (up from about 125,000 records behind the 2021 edition), covering 643 unique CWEs mapped to those CVEs (up from 241 in 2021), extracted through OWASP Dependency Check. Contributed testing data came from 12 named organizations plus several anonymous donors across more than 2.8 million tested applications, and this time OWASP asked for data "with no restriction on CWEs," which put 589 CWEs into the analysis. From that pool, 248 CWEs were grouped into the 10 final categories, averaging 25 CWEs per category, ranging from 5–6 (A03, A09) up to a cap of 40 in Broken Access Control. Risk ranking combines incidence rate, exploit score, impact score and total occurrence count into a single weighted formula per OWASP's own published methodology.
The introduction is explicit that frequency inside one application does not count: "Whether an application has four instances of a CWE or 4,000 instances is not part of the calculation." Exploit and impact scores come from CVSS v2 and v3 sub-scores; CVSS v4.0 was not used because its algorithm "no longer easily provides the Exploit or Impact scores."
The qualitative side is OWASP's community survey: per the project's own introduction, "We ranked 12 categories based on the data contributed, and allowed two to be promoted or highlighted by responses from the community survey," and elsewhere it says "we only pick eight of the ten categories from the data." The introduction ties two categories to the survey: Software Supply Chain Failures, "overwhelmingly voted a top concern," and Security Logging & Alerting Failures, which "was again voted into a position in the list from the community survey participants." OWASP frames this explicitly as a way to surface risks that automated testing underrepresents, such as a rare-but-catastrophic supply-chain compromise.
How should you read a category's score table?
Each category page opens with the same nine figures. What they mean, per the methodology page and introduction:
| Field | Meaning |
|---|---|
| CWEs Mapped | How many CWEs OWASP grouped into the category |
| Max / Avg Incidence Rate | Share of tested applications with at least one of those CWEs (highest single CWE / mean across them) |
| Max / Avg Coverage | Share of applications that were tested for those CWEs |
| Avg Weighted Exploit / Impact | CVSS v2/v3 exploitability and impact sub-scores of CVEs with those CWEs, weighted by the v3 share |
| Total Occurrences | Number of occurrences of those CWEs OWASP counted in the contributed data |
| Total CVEs | CVEs in NVD that carry one of the CWEs |
Read A03 and A09 with this in mind: low coverage and few CVEs mean the testing tools rarely look for these weaknesses, not that they are rare. A09 lists only 723 CVEs, and OWASP says it "will always be underrepresented in the data."
How does each category show up in a connected device?
The Top 10 is written for web applications, but a router's admin page, a camera's embedded web server and a device's cloud API are web applications. Each row below uses a CWE that OWASP lists in that category:
| Category | Typical device case | CWE OWASP lists |
|---|---|---|
| A01 Broken Access Control | Admin CGI endpoint served without a session check; "test this URL" field that lets the device fetch internal addresses | CWE-862, CWE-918 |
| A02 Security Misconfiguration | Telnet or a debug web page left on in production builds; A02 lists "Default accounts and their passwords are still enabled and unchanged" | CWE-16 |
| A03 Software Supply Chain Failures | Firmware shipping an outdated BusyBox or OpenSSL from a chip-vendor SDK; a component that cannot be updated in the field | CWE-1395, CWE-1329 |
| A04 Cryptographic Failures | One TLS private key or AES key in every unit's image | CWE-321, CWE-338 |
| A05 Injection | Ping or traceroute diagnostics page passing a hostname into a shell | CWE-78 |
| A06 Insecure Design | Wi-Fi and cloud credentials stored in plain text in flash | CWE-256 |
| A07 Authentication Failures | Shared hard-coded support password; A07 says "Do not ship or deploy with any default credentials" | CWE-798, CWE-1392 |
| A08 Software or Data Integrity Failures | Update handler that flashes an image without a signature check | CWE-829 |
| A09 Security Logging & Alerting Failures | No record or alert for repeated failed logins on the web UI | CWE-778 |
| A10 Mishandling of Exceptional Conditions | Parser crash on a malformed packet; a service that "fails open" after an error | CWE-476, CWE-636 |
A08 is the one category where OWASP names firmware itself: its unsigned-update scenario says "many home routers, set-top boxes, device firmware, and others do not verify updates via signed firmware," and adds that often "there is no mechanism to remediate other than to fix in a future version." For the device side of that control, see secure boot and hardware root of trust. For the weakness-level view behind these CWEs, see the CWE Top 25.
Worked example: how would a team triage a router's web UI against the list?
Synthetic example, not a customer case. A router maker reviews its web admin interface and update path before a new firmware release:
Inventory the surfaceList every CGI handler, API route and upload endpoint, plus the update and recovery paths. Each becomes a row to test against A01, A05 and A08. Check access control first (A01)Request each admin route without a session and with a low-privilege session. Two diagnostics routes answer without a session: logged as CWE-862. Test injection (A05)The traceroute field accepts ;and runs it in a shell: CWE-78. Combined with the A01 finding, it is reachable without login, so it goes to the top of the list.Check the update path (A08)The image header carries a CRC but no signature; a modified image installs. This needs a bootloader change and key management, so it is scheduled for the next hardware-supported release. Pull the component list (A03)An SBOM of the image shows an OpenSSL branch out of support; its CVEs are matched against NVD and CISA KEV, and severity is checked in a CVSS calculator. Record the restDefault "admin/admin" on first boot (A07), no log of failed logins (A09) and a crash on an oversized HTTP header (A10) each get a ticket with the OWASP category and CWE.
The order follows exploitability on this device, not the A01–A10 numbering. For ranking component CVEs, see KEV and EPSS prioritization.
What are the common pitfalls?
- Reading the rank as a fix order. OWASP ranks prevalence and severity across 2.8 million applications; one device's exposure can make A08 or A07 more urgent than A01.
- Treating the Top 10 as a test standard. OWASP calls it an awareness document. For verifiable requirements use OWASP ASVS; for devices, the OWASP ISVS.
- Mixing 2021 and 2025 codes. A06:2021 and A06:2025 are different categories; record the year with every code.
- Scanning only the web UI. A03 and A08 live in the build and update path, which a web scanner never sees.
- Skipping A10 because it is new. NULL-pointer and fail-open bugs in C parsers are everyday firmware defects.
What license is the Top 10 under?
The OWASP/Top10 GitHub repository — the source of the published document — carries a LICENSE file identifying the Creative Commons Attribution-ShareAlike 4.0 International Public License, meaning the Top 10 text can be reused and adapted with attribution, provided derivative works carry an equivalent CC license.
Where does this fit for a device or firmware maker?
The OWASP Top 10 is scoped to web application risks specifically, rather than embedded firmware or OT protocols — for those, see our guides on the OWASP IoT Top 10 and OWASP ASVS for a more rigorous, testable verification standard, and the cluster's OWASP Firmware Security Testing Methodology. But most connected devices now ship a web-based management UI or cloud API, which makes the 2025 list directly relevant: A01 Broken Access Control and A03 Software Supply Chain Failures in particular map closely onto the device-update and component-provenance questions covered in our firmware SBOM and CSAF/VEX for device makers guides, and onto the API-specific risks in our OWASP API Security Top 10 guide for devices exposing a dedicated API rather than a browser UI.
SoftWSP covers the component side of A03: it extracts an SBOM from the firmware image as CycloneDX 1.6 JSON, matches the components against NVD, CISA KEV, FIRST EPSS and CISA CSAF ICS advisories, and drafts a cited report that an engineer reviews; watchlist alerts flag new CVEs by CVE id or vendor-product. Access control, injection and error handling in the web UI need separate testing. See device firmware security.
This guide summarises OWASP's own public Top 10:2025 publication; it does not constitute legal or compliance advice.
Put this guide to work
Frequently asked questions
What is the OWASP Top 10?
A standard awareness document published by the Open Worldwide Application Security Project (OWASP) that represents what OWASP describes as "a broad consensus about the most critical security risks to web applications." The current edition, OWASP Top 10:2025, is the project's eighth installment since the first list in 2003 (prior editions: 2004, 2007, 2010, 2013, 2017, 2021); it was announced as a release candidate at the OWASP Global AppSec Conference in November 2025 and OWASP's own repository README now describes it as released and "Final."
What are the 10 categories in the 2025 edition?
A01:2025 Broken Access Control, A02:2025 Security Misconfiguration, A03:2025 Software Supply Chain Failures, A04:2025 Cryptographic Failures, A05:2025 Injection, A06:2025 Insecure Design, A07:2025 Authentication Failures, A08:2025 Software or Data Integrity Failures, A09:2025 Security Logging & Alerting Failures, and A10:2025 Mishandling of Exceptional Conditions — per OWASP's own published category list.
What changed from the 2021 list?
OWASP's own introduction states there are "two new categories and one consolidation" for 2025. Software Supply Chain Failures (A03:2025) is new, expanding what 2021 called A06: Vulnerable and Outdated Components into a broader supply-chain-compromise category; Mishandling of Exceptional Conditions (A10:2025) is also new. The consolidation: 2021's standalone A10 (Server-Side Request Forgery) was folded into A01:2025 Broken Access Control. Three categories were also renamed — 2021's "Identification and Authentication Failures" became "Authentication Failures," "Security Logging and Monitoring Failures" became "Security Logging & Alerting Failures," and "Software and Data Integrity Failures" became "Software or Data Integrity Failures" — and several categories moved rank, most notably Security Misconfiguration rising from 5th to 2nd.
How does OWASP decide the ranking and what license is the list under?
Per OWASP's own methodology page, eight of the ten 2025 categories were selected by ranking candidate categories against contributed vulnerability-testing data (incidence rate, exploit and impact scores drawn from roughly 175,000 CVE records mapped to CWEs, contributed by 12 named organizations plus anonymous donors across more than 2.8 million tested applications), while two categories were "promoted or highlighted by responses from the community survey" to capture risks the automated data does not measure well — Software Supply Chain Failures was the #1 community-survey pick despite having the fewest occurrences in the contributed data and only 11 related CVEs, reflecting that its data-driven score alone would have ranked it lower. The OWASP/Top10 GitHub repository is published under the Creative Commons Attribution-ShareAlike 4.0 International license.
Sources
- OWASP Top 10:2025 — official project page (top10.owasp.org)
- OWASP Top 10 GitHub repository — 2025 edition source
- OWASP Top 10:2025 — 0x00 Introduction (what changed from 2021)
- OWASP Top 10:2025 — 0x02 What Are Application Security Risks (methodology, data volume)
- OWASP Top 10:2025 — A01 Broken Access Control (category detail, CWE/incidence data)
- OWASP Top 10:2025 — A03 Software Supply Chain Failures (category detail)
- OWASP Top 10:2025 — A02 Security Misconfiguration
- OWASP Top 10:2025 — A07 Authentication Failures
- OWASP Top 10:2025 — A08 Software or Data Integrity Failures (unsigned firmware scenario)
- OWASP Top 10:2025 — A09 Security Logging & Alerting Failures
- OWASP Top 10:2025 — A10 Mishandling of Exceptional Conditions
- OWASP Top 10 GitHub repository — README (2025 "RELEASED", 2021 "SUPERSEDED")
- OWASP Top 10 GitHub commit f7439d4 (24 December 2025): Release Candidate styling removed as 2025 goes live
- OWASP Top 10 GitHub repository — LICENSE (CC BY-SA 4.0)
About this guide
SoftWSP Firmware Security Desk
Covers the Firmware vulnerabilities & fixes cluster: 129 guides.

In this cluster
Questions about this page?
