WordPress Core: unauthenticated attacker can run code on the server
Someone on the internet without a login could trick WordPress into loading a PHP file it should not, which can let them take control of the website server. Attacks on this flaw are confirmed.
| Severity | CriticalVulnerability |
|---|---|
| Being exploited? | YesConfirmed by CISA KEV · 25 Sept 202 |
| Fix available? | UnclearNo fixed version published in the sources yet |
| How sure is this? | Official sourceBacked by an official source |
How the attack works simplified
- 1Unauthenticated attackerNo login needed to send requests
- 2Crafted page requestManipulates the page-template lookup
- 3WordPress includes fileLoads a chosen local .php file outside theme dirs
- 4Code executesIncluded file runs on the server
- ResultAttacker controls the web server
What happened
A remote file inclusion flaw in WordPress Core lets an unauthenticated attacker make page-template resolution include a chosen readable local .php file outside the active theme directories, leading to remote code execution.
Are you affected?
You are affected if your site runs WordPress Core and is reachable from the internet.
WordPress Core (exact versions not stated in the provided material)
What to do
- Check CISA guidance and the WordPress vendor instructions for mitigations and updated versions, then apply them immediately
- Confirm which of your WordPress sites are exposed to the internet and prioritize them
- If no mitigation is available, follow CISA BOD 26-04 guidance or take the product offline
- Review servers for signs of compromise using CISA's Forensics Triage Requirements
Timeline
- 25 Sept 2026, 00:00 UTCPublicly disclosed
- 25 Sept 2026, 00:00 UTCReported by CISA KEV · CISA KEV
Sources 1 report, oldest first
Official vendor, government CERT or advisory database
- CISA KEV25 Sept 2026, 00:00 UTC
Technical details
- ID
- CVE-2026-87902
- EPSS
- 96% chance of exploitation within 30 days
- Disclosed
- 25 Sept 2026, 00:00 UTC
- On Zero Radar
- before the radar started
- Last change
- 25 Oct 2026, 13:32 UTC
The plain-language parts and the attack diagram were written by an AI model (anthropic/claude-haiku-5.5) from the sources above. Facts like versions, exploitation and dates come from the sources themselves. How Zero Radar works