WordPress 7.0.3 Patches High Severity XSS Vulnerability Across 12 Security Flaws

WordPress released security update 7.0.3 on August 6, 2026, patching twelve vulnerabilities in its core software — including a high severity login screen flaw rated 8.9 out of 10 that could lead to remote code execution.

The sheer volume of vulnerabilities in a single release marks an unusual moment for the platform that powers a significant portion of the web. WordPress has historically seen relatively few core vulnerabilities but has experienced a notable cluster of discoveries recently — a trend security researchers link directly to the growing use of AI-powered penetration testing tools. For the millions of website owners and administrators running WordPress, this update demands immediate attention.

---

What Makes This Release Different

The vulnerabilities patched in 7.0.3 were reported by a striking mix of contributors including Anthropic AI, pwn_ai, and Aikido Security — signaling how artificial intelligence is reshaping the landscape of cybersecurity research. AI tools are now uncovering flaws that human researchers previously missed, and doing so at a scale and speed that simply wasn't possible before.

Oliver Sild of Patchstack — a WordPress security platform — captured the mood in the security community on X:

"Weeks ago when #WP2Shell dropped and OpenAI Sol Ultra took most of the credit — all I thought about was how all the other labs and AI-pentest companies will rush to prove they can find something in WordPress core too. Well, today, 3 weeks later — we have the next WordPress core version release with not one but 12 vulnerabilities being patched."

WordPress's official announcement offered minimal context about severity levels, leaving many site owners without a clear picture of urgency. That lack of transparency is a recurring frustration for administrators who rely on those disclosures to prioritize patching decisions. Understanding why cybersecurity matters for businesses of all sizes has never been more relevant — the pace of AI-driven discovery means the window between vulnerability identification and active exploitation is narrowing.

---

Breaking Down the 12 Vulnerabilities

Of the twelve flaws addressed in this release, three stand out as particularly serious for site owners and security teams.

Pre-Auth Reflected XSS on the Login Screen (Rated 8.9/10 — High)

This is the most critical vulnerability in the batch. Described as a pre-authentication flaw, it means an attacker does not need a WordPress account to initiate an attack. Via a specially crafted malicious third-party website, an attacker can exploit this vulnerability and — under certain conditions — escalate it to remote code execution (RCE).

The Open Worldwide Application Security Project (OWASP) explains the broader risk: "An attacker can use XSS to send a malicious script to an unsuspecting user. The end user's browser has no way to know that the script should not be trusted and will execute the script."

Exploitation does require social engineering — meaning a legitimate user must be tricked into taking a specific action — which limits the likelihood of mass exploitation. WordPress's official GitHub security repository confirms the fix has been backported to all WordPress branches dating back to version 4.7.

Server-Side Request Forgery (SSRF) in URL Validation

This vulnerability allows requests to link-local IP ranges — internal server addresses not meant to be accessible from outside. While WordPress provided no detailed description of this flaw, its implications could include exposure of sensitive server-side information. The absence of a severity rating makes it difficult to fully assess the risk at this stage.

Privilege Escalation on Multisite Networks

Sites running WordPress in multisite mode with user registration enabled are exposed to a flaw that allows an unauthorized user to create a new site within the network. This carries particular risk for universities, enterprise networks, and organizations managing multiple sites under one installation.

The remaining nine vulnerabilities — while lower in individual severity — include:

• Multiple stored XSS issues affecting contributors and authors
• An information disclosure flaw in the Latest Comments block
• An email confirmation bypass
• CSS injection and post slug enumeration issues
• A disclosure of notes in comment feeds

For site owners who want a deeper understanding of the attack surfaces these flaws represent, a thorough review of how website security works and where vulnerabilities commonly emerge provides essential context for evaluating your own exposure.

---

How Worried Should Site Owners Actually Be?

Assessing the Real-World Risk

Sild offered a measured assessment when speaking to Search Engine Journal directly: "We're tracking whether it's getting exploited but looks like the social engineering bit on the XSS that could lead to RCE is most likely not going to get much attention from the hackers."

That is somewhat reassuring — but it does not eliminate the need to act. Sild also noted that unlike the WP2Shell vulnerability from three weeks prior — which was considered mass-exploitable — none of the flaws in this release appear to carry that level of immediate widespread risk.

WordPress's auto-update mechanism is working in site owners' favor. Sild confirmed that "WordPress is auto-updating fast" and that Patchstack customers had already received mitigation rules at the time of disclosure.

For sites not set to auto-update, the window of exposure remains open until a manual update is applied.

The backport of the critical XSS fix to all versions back to 4.7 demonstrates that WordPress is taking the login screen vulnerability seriously and attempting to protect as broad a user base as possible — regardless of which version they're running.

The Bigger Picture: AI Is Changing the Threat Landscape

What this release makes unmistakably clear is that the security environment around WordPress is shifting. The involvement of AI systems in discovering these twelve vulnerabilities is not incidental — it reflects a structural change in how security research is conducted. AI-powered penetration testing tools can probe codebases continuously, systematically, and at a depth that no human team can match at scale.

This has a direct consequence for site owners: the gap between vulnerability discovery and active exploitation is shrinking. Security teams are finding flaws faster, but so are malicious actors using the same tools. The sites most at risk are those operating on delayed or manual update cycles — left exposed during the interval between disclosure and patching. Keeping up with WordPress website security best practices is no longer an optional extra for serious site owners — it's a baseline requirement.

What Site Owners Should Do Now

The practical steps from this release are clear and time-sensitive.

1. Verify your WordPress installation has updated to version 7.0.3. Check your dashboard under Updates if auto-updates are not confirmed. Do not assume the update has applied without verifying.
2. If you manage a multisite network with user registration enabled, review your user roles and site creation permissions immediately. The privilege escalation flaw poses a direct risk to your network's integrity and should be treated as a priority regardless of your overall risk tolerance.
3. Consider implementing a security plugin or web application firewall — such as Patchstack or a comparable service — that provides real-time virtual patching. This is particularly valuable in the gap between a vulnerability disclosure and a completed site update, when exposure is at its highest.

The broader lesson from this release is one worth absorbing for long-term site management: AI-driven security research is accelerating the pace of vulnerability discovery, and site owners who rely on delayed or manual update processes are increasingly exposed in that gap between disclosure and patch deployment. The question is no longer whether your platform will face new vulnerabilities — it's whether your update and monitoring processes are fast enough to respond when they arrive.