Security1 distinct publisher2 min readPublished
Wordfence blocked more than 250,000 attempts against Super Forms and 190,000 against Elementor Pro. The Super Forms campaign has been running since July 14, so unpatched sites need a look through their uploads directories.
The Watch · Security desk

Compiled by The WatchSomething wrong?How this is made
The Super Forms request is one HTTP POST to /wp-admin/admin-ajax.php against the super_submit_form endpoint, with the file field carrying a base64 payload and an attacker-chosen filename [5]. The blob is prefixed data:image/gif;base64 and decodes to PHP, because the plugin does not check the type before writing [1][6]. What lands is Mushr00w_upl.php, a web shell whose function is to accept further uploads [6]. Eviction order follows from that: deleting the first file achieves nothing if a second stage went in behind it. The sampled request also carries form_id=2 and an sf_nonce value, which makes it cheap to grep in access logs [18].
Elementor Pro fails differently. The File Upload field is submitted as an array, empty first element and PHP payload second, and Wordfence says that structure is what triggers the validation bypass [10]. The file is written to /wp-content/uploads/elementor/forms/ under a randomly generated name with the attacker's .php extension, then requested directly [11]. Two hunt targets, then: .php files in that directory, and admin-ajax.php POSTs containing super_submit_form.
The two block counts add to more than 440,000 [4], and they are what the label says, attempts Wordfence stopped on sites it protects [3]. The reporting carries no figure for successful compromises and no release date for either patch [17]. High volume tells you the scanning is industrialised, not how many shells are resident.
Elementor Pro exploitation opens on August 19, 2026 [8], one day after the Super Forms activity peaked [9], against a bug Patchstack had documented the month before [13]. The primitive is identical in both campaigns: unauthenticated write of a PHP file into a path the web server executes. The attention moves to whichever plugin was most recently written up.
Configuration decides the Elementor half of the triage. A site below 4.2.2 is only exploitable if it has at least one published page with a Form widget containing a File Upload field [12]. No equivalent precondition is stated for Super Forms [16], where an active plugin is the whole requirement. A version report therefore ranks the Elementor estate badly, and a page inventory ranks it properly.
CVE-2026-32475 is scored 9.0 in one place and 9.8 in another [2]. The spread has no operational content: unauthenticated file write into an executable directory sits at the front of the queue on either number.
Wordfence lists ten source IPs on the Super Forms side [15]. Blocking them is housekeeping. The reachable outcome for any hit that got through is an administrator account and control of the site, with the database in between [14].
Ranked by verification strength, evidence, and original report placement.
CVE-2026-14894 is a missing file type validation vulnerability in the Super Forms - Drag & Drop Form Builder WordPress plugin, CVSS 9.8, allowing unauthenticated attackers to upload files of any type including executable PHP, leading to remote code execution; fixed in version 6.3.314.
CVE-2026-32475 is a vulnerability in Elementor Pro, given a CVSS score of 9.0/9.8, that allows unauthenticated attackers to upload files of any type including executable PHP, leading to remote code execution; fixed in version 4.2.2.
In a pair of reports published this week, Wordfence said it had blocked over 250,000 exploit attempts targeting CVE-2026-14894 and over 190,000 targeting CVE-2026-32475.
In attacks exploiting CVE-2026-14894, threat actors issue an HTTP POST request to /wp-admin/admin-ajax.php using the super_submit_form endpoint, containing a file field with a Base64-encoded PHP payload and an attacker-controlled file name.
The uploaded file is prefixed with the data:image/gif;base64 content type but is a PHP file-uploader web shell named Mushr00w_upl.php, which acts as a conduit to upload additional payloads to the site.
The malicious activity against Super Forms began on July 14, 2026, and scaled to a peak of more than 40,000 exploit requests on August 18, 2026.
Distinct publishers with included, body-backed reporting in this cluster.
1 article · September 4, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
security
Attackers started dropping webshells through Elementor Pro forms on patch day1 distinct publisher
security
Two loops, one blocklist bypass: Elementor Pro's upload field becomes unauthenticated RCE4 distinct publishers
security
Two miniOrange SAML bugs under attack, and 30,000 paid installs were never told5 distinct publishers
security
GiveWP issues accounts to unauthenticated attackers on sites where registration is off3 distinct publishers
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One vendor's telemetry, unusually checkable
Every number traces back to two Wordfence reports, relayed by The Hacker News without a second network's view or a word from either plugin vendor. What lifts it above ordinary single-sourcing is how falsifiable the detail is: the sample POST decodes to exactly the uploader shell described, the endpoint and directory paths can be checked against a live install, and twenty source addresses are named. Anyone running these plugins can confirm or refute the mechanics from their own logs — but not the 440,000 figure, which only Wordfence can see.
Attacker adoption is the measured thing
Read adoption here as uptake by the attackers, and it is substantial: seven weeks of sustained activity against Super Forms with a 40,000-request day, a second campaign opening against Elementor Pro the day after that peak, and distinct address sets behind each. Missing is the other half — what actually landed. Wordfence counts requests it stopped, so intent is measured and success is not, and no number of compromised sites appears anywhere in this reporting.
440,000 attempts, zero confirmed victims
The headline number is a count of things that did not work. Wordfence blocked them; that is what a firewall vendor's telemetry records. The prose is honest about this — "blocked over 250,000" — but the aggregate travels as though it measured damage, and nothing in the reporting says a single site fell. Pushing the same direction: the Elementor Pro flaw only bites sites with a published Form widget carrying a File Upload field, a qualifier that sits in one clause well below a 440,000-attempt headline. The overstatement is modest and structural rather than deliberate.
The threat data is also the product demo
Wordfence sells WordPress firewalling, and the proof that these campaigns are dangerous is a count of attacks its own firewall stopped — a genuinely useful disclosure that also happens to be the best possible advertisement for it. Patchstack, credited with disclosing the Elementor Pro flaw a month earlier, competes in the same market on the same currency. None of this makes the mechanics wrong; the exploit request either decodes as described or it does not. It does mean the one figure nobody else can verify is the one that most flatters the party supplying it.
Act on the mechanics, hold the totals loosely
Confidence splits by what kind of fact you need. The bug classes, the fixed versions, the request structures and the file paths are specific, internally consistent and testable, and they are what a response actually runs on. The scale numbers, the campaign start dates and the source address lists come from one vendor through one outlet with no second reading, and the missing patch dates mean nobody can say how long sites sat exposed. Enough to patch and hunt on; not enough to quote the totals as settled.