CVE-2025-8570 is a privilege escalation vulnerability in the BeyondCart Connector plugin for WordPress. The flaw is caused by improper JWT secret management together with insufficient authorization logic in the plugin’s handling of the WordPress determine_current_user filter. In affected versions, the plugin accepts JWTs that can be forged because the signing secret is hard-coded or otherwise predictable, allowing an attacker to generate valid tokens with arbitrary identity claims. By presenting a crafted token, an unauthenticated attacker can cause the plugin to treat them as any chosen WordPress user, including an administrator.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
1 valid exploit after Mallory filtered fakes, detection scripts, and README-only repos (1 hidden).
This repository contains a Python exploit script (CVE-2025-8570.py) targeting a critical privilege escalation vulnerability (CVE-2025-8570) in the BeyondCart Connector WordPress plugin (versions 1.4.2 through 2.1.0). The vulnerability arises from improper JWT secret management, allowing unauthenticated attackers to generate valid JWT tokens and impersonate any user, including administrators. The script automates the exploitation process by generating or accepting a JWT, then sending a crafted PUT request to the vulnerable REST API endpoint (/wp-json/grind-mobile-app/v1/customers/1) to update administrator account details (email, username, password, etc.). The repository includes a README with detailed usage instructions, a requirements.txt listing dependencies (requests, pyjwt), and a custom license. The exploit is operational, providing direct privilege escalation and user data modification on vulnerable targets. No hardcoded IPs or domains are present; the script requires the user to supply the target URL.
5 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.