WordPress has rolled out an automated, AI-driven security review that screens every plugin release before it reaches the WordPress.org update API, adding a critical checkpoint to a distribution pipeline that had previously lacked one.
The move follows a real-world incident in which a backdoor was slipped into an update for a plugin with roughly 20,000 active installations, exposing a dangerous gap between when a plugin release is committed and when it reaches millions of live sites.
On July 28, 2026, a malicious commit was pushed into a plugin release already sitting inside WordPress.org’s mandatory cooldown window. The automated review system flagged the update with a high security score, and because the release had not yet cleared its cooldown period, it was never distributed through the update API to end users.
WordPress Uses AI to Stop Malicious Plugin
The Plugins Team pulled the compromised plugin from the directory just 26 minutes after security firm Wordfence alerted them, but the incident revealed a structural weakness: detection alone wasn’t enough if a human still had to intervene to stop distribution.
That realization directly led to building an automated blocking mechanism that acts without waiting on the availability of a Plugins Team reviewer.
Since June 5, 2026, every plugin and theme release has passed through a mandatory six-hour cooldown before becoming available through the WordPress.org update API, including one-click updates triggered from the WordPress dashboard.
During this window, multiple AI models now analyze the code changes in each release alongside Jetpack Scan, WordPress’s existing malware and vulnerability scanning engine.
These tools cross-check and consolidate their outputs into a single security score, where a higher score signals greater potential risk. Running several detection systems in parallel and comparing their findings helps keep the process accurate and reduces false positives, though WordPress.org acknowledges the system is not entirely error-free.
Releases that cross a defined risk threshold are blocked automatically the moment the review completes, and every committer on the plugin receives an email detailing the specific findings that triggered the block.
Releases that fall below the threshold continue through the standard cooldown and are distributed normally. Importantly, WordPress.org has clarified that a high score does not automatically imply malicious intent; an unintentional coding flaw or vulnerability can register the same elevated score as deliberately planted malware, since the system measures risk exposure rather than the developer’s intent.
For now, notification emails are sent only when a release is actually blocked, so plugin authors who don’t hear anything can assume their update passed review without issue.
When a release is blocked, it simply will not be pushed through the update API until the underlying issues are resolved. WordPress.org recommends that authors first review the specific findings included in the notification email, since these outline exactly what triggered the flag.
The fastest path to resolution is to fix the identified issues and publish a new release; once the revised version scores below the blocking threshold, it re-enters the normal cooldown and distribution pipeline.
Authors who believe a finding is a false positive can contact the Plugins Team directly, though the team has cautioned that publishing a corrected release is almost always faster than waiting for a manual appeal review, given the high volume of submissions they process.
WordPress powers a significant share of the web, and plugins represent one of its most persistent attack surfaces, since a single compromised update can silently propagate to every site running that extension.
By embedding AI-based scanning directly into the update pipeline rather than relying solely on post-release detection, WordPress.org is shifting toward a proactive, supply-chain-style defense model similar to approaches increasingly adopted across software distribution platforms.
The Plugins Team has signaled that the detection thresholds and models will continue to evolve as more data is collected, and it has specifically asked developers to report false positives to help refine the system’s accuracy over time.
For an ecosystem where plugin updates auto-deploy to millions of sites with minimal friction, this automated gatekeeping represents one of the more consequential security changes to WordPress’s infrastructure in recent years.
Learn 7 Metric-Gated AI SOC Deployment Phases – Download Free AI SOC Deployment Playbook 2026.
