WordPress AI blocks malicious plugin updates
WordPress now uses AI to scan every plugin update before it reaches users, stopping a backdoor attack that slipped past human checks in July 2026. The system automatically blocks suspicious releases without waiting for manual review.
- Severity
- Not scoredNo CVSS score recorded
- Affects
- WordPress
How it works
- A malicious developer pushed a backdoored plugin update to WordPress.org on July 28, 2026.
- The update was already in the mandatory six-hour cooldown period, but Wordfence flagged it as high-risk.
- WordPress's Plugins Team removed it manually within 26 minutes.
- This incident showed that human review alone was too slow to stop malicious updates from reaching users.
- WordPress then added AI-powered automated blocking to reject suspicious releases before they reach the update API, without needing a human to act.
What to do
This prevents malicious updates from being installed, but you should still review plugin updates manually to ensure they are safe.
However, always review plugin updates manually to ensure they are safe and trustworthy.
Technical details
WordPress introduced an AI-powered security review system in June 2026 to automatically block malicious plugin updates before they reach users. The system analyzes code changes during a mandatory six-hour cooldown period, using AI models alongside Jetpack Scan to detect risks like backdoors or vulnerabilities. If a release exceeds a predefined risk threshold, it is blocked immediately, and the plugin's developers are notified with details of the findings.
This follows an incident on July 28, 2026, where a malicious update for a plugin with ~20,000 active installations was flagged by Wordfence but required manual intervention to prevent distribution. The new system eliminates this dependency on human review.