AI Is Finding WordPress Flaws Faster Than Ever — Anthropic and OpenAI Credited in 7.1.3 Release Notes
Something important has changed about who finds WordPress security flaws. Three of the seven fixes in WordPress 7.1.3 are credited to Anthropic, the AI company — plus two more in 7.1.1 and a collaboration between Trail of Bits and OpenAI on the comments XSS flaw. The age of AI WordPress flaws discovery has arrived, and the numbers are startling: WordPress’s bug bounty program jumped from reports “in the dozens” per month for nine years to 450 in July 2026. For site owners, the practical takeaway is simple — both sides now work faster, and patch windows are measured in hours.
What Happened: AI WordPress Flaws Go Mainstream
The WordPress 7.1.3 release notes read like a roster of the AI era. Anthropic is credited with three vulnerabilities: a denial-of-service bug in the WP_Http::make_absolute_url() method, a second-order SQL injection in WordPress’s WXR export, and a capability check that let author-level users make posts sticky. Every WordPress security release since July has credited AI companies or AI-assisted discovery, with 7.1.2 the lone exception. Anthropic also reported two of the eleven fixes in WordPress 7.1.1 the previous month.
Trail of Bits, in collaboration with OpenAI, reported the stored cross-site scripting flaw on the Comments administration page. The remaining credits went to humans: Ananda Dhakal of Patchstack found the unauthenticated comment disclosure, university researchers flagged the Imgur embed XSS, and the WordPress security team itself caught the hook-collision issue.
Analysis of the AI-built attacks is what really turns heads. One AI-constructed attack on WordPress core needed no login, took about ten hours to develop, and cost roughly $25 in compute. A separate research pipeline — pairing AI static analysis with automated verification — surfaced more than 300 critical zero-days across the WordPress plugin ecosystem in about 72 hours of scanning, with every finding manually verified before disclosure. The AI WordPress flaws pipeline is not a curiosity anymore; it is the dominant source of new findings.
The scale extends far beyond WordPress. Anthropic’s broader scanning program surfaced more than 29,000 candidate vulnerabilities across widely used open-source projects in six months — though outside firms had confirmed only about 6,000 as valid and just 516 had been patched upstream by early October. The discovery engine is running far ahead of the patching engine, and that gap is the story.
Who Is Affected by the AI WordPress Flaws Surge
Every WordPress site owner is affected, because the surge in AI WordPress flaws discovery changes the timeline of every future disclosure. When finding a vulnerability costs $25 and ten hours instead of weeks of expert labor, the supply of disclosed flaws goes up — and Patchstack’s 2026 report counted 11,334 new ecosystem vulnerabilities in 2025, with 91 percent in plugins. That number will not be going down.
The double-edged nature of the shift is the critical point. The same automation that lets a researcher find a critical WordPress flaw for $25 is available to anyone, including people who do not report what they find. Once a fix is published, the code changes are visible to everyone — attackers included. The WordPress 7.1.2 release on September 22 is the textbook case: it fixed a critical flaw in how page templates were loaded that could let anonymous users execute code on some server configurations, and Patchstack’s firewall network saw the first exploitation attempts at 11:49 AM UTC that same day. Within four hours, attack traffic had grown to more than ten times the initial volume, with scanning tools circulating. A business owner who went to sleep with an unpatched site woke up to a compromised one.
Plugin developers face a parallel squeeze. AI now writes plugins fast — “vibe coding,” where developers ship LLM-generated plugin code they cannot actually audit — and skips the boring security parts: escaping, capability checks, nonce validation. One agency reported finding 100 distinct security issues in a single vibe-coded plugin. The two things that used to protect a small plugin, obscurity and time, are both gone.
Even the institutions are buckling. On October 1, 2026, Google temporarily suspended product vulnerability submissions under its open-source bug bounty program, citing “a significant rise in automated submissions, the vast majority of which are not valid.” Anthropic responded by launching a free AI vulnerability scanner that requires every report to include a proof-of-concept reproducer and a suggested patch. The triage burden of the AI era is a problem the whole industry is still solving.
Why the AI WordPress Flaws Shift Matters
The shift matters because it rewrites the economics of both attack and defense. For nine years, WordPress core received a few dozen bounty reports a month — a pace the security team could absorb. At 450 reports in a month, triage itself becomes the bottleneck. Poland’s government cybersecurity office issued a formal advisory over WordPress core flaws patched back in July — two vulnerabilities that, chained together, could let an attacker inject malicious database commands and ultimately run code — classifying them as a threat to public infrastructure. When nation-state advisories start tracking WordPress patch notes, the stakes are no longer theoretical.
For defenders, the news is genuinely good: AI-assisted research is surfacing longstanding bugs that sat quietly in code for years, and responsible disclosure means they get fixed. WordPress bug bounty payouts topped $90,000 in 90 days, and every credited AI finding in 7.1.3 was privately reported and patched before release. The software millions of businesses depend on is undergoing heightened scrutiny.
For site owners, though, the math is unforgiving. Patchstack measured the weighted-median time from public disclosure to mass exploitation at roughly five hours. The standard advice — keep your plugins updated — assumes you have a window to react. Five hours is not a window. The only reliable defense is to shrink the gap between a patch’s release and its installation to as close to zero as possible: automatic updates, monitored sites, and a plan that does not depend on anyone remembering to click a button.
What to Do Now in the Era of AI WordPress Flaws
The AI WordPress flaws surge rewards owners who automate and punishes owners who procrastinate. Here is the playbook:
- Enable automatic updates for WordPress core, plugins, and themes. This is now the single highest-leverage security action. With disclosure-to-exploitation measured in hours, manual update cycles cannot keep up.
- Verify auto-updates are actually working. Check Dashboard → Updates and Site Health. Host-level filters or misconfigurations can silently disable background updates.
- Put a security scanner or firewall in front of your site. When patches cannot be applied instantly, virtual patching from a web application firewall buys time against known attack patterns.
- Reduce your attack surface. Delete unused plugins and themes — every installed component is code an AI-assisted attacker can now probe cheaply. Fewer plugins means fewer AI WordPress flaws to worry about.
- Monitor for unexpected changes. File-integrity monitoring and uptime alerts catch the compromises that slip through, so a breach becomes an incident you discover in minutes, not months.
- Keep backups current and tested. When an AI-discovered flaw becomes an AI-exploited flaw on your site, a clean, tested backup is the difference between a bad afternoon and a lost business.
AI WordPress Flaws: Frequently Asked Questions
Are AI companies really finding WordPress vulnerabilities?
Yes. Anthropic is credited with three of the seven fixes in WordPress 7.1.3 and two in 7.1.1, and Trail of Bits collaborated with OpenAI on the comments XSS fix. AI-assisted discovery has been credited in every WordPress security release since July 2026.
How fast is AI finding WordPress flaws?
WordPress’s bug bounty program went from dozens of reports per month for nine years to 450 in July 2026. One AI-built attack on core needed no login, took about ten hours, and cost roughly $25. A separate pipeline found 300+ verified critical zero-days in plugin code in about 72 hours.
Does AI finding flaws help or hurt site owners?
Both. Responsibly disclosed AI WordPress flaws get patched before release — genuinely good news. But the same tools are available to attackers, patches are public immediately, and exploitation now follows disclosure within hours. The net effect: you must patch faster than ever.
What is the single best defense against AI-discovered vulnerabilities?
Automatic updates. With a median of roughly five hours from public disclosure to mass exploitation, manual update routines cannot keep pace. Auto-updates plus monitoring shrink your exposure window to the minimum possible.
Should I worry about AI-written plugins?
Be cautious. “Vibe-coded” plugins — LLM-generated code the developer cannot audit — are shipping with serious flaws: one agency found 100 distinct security issues in a single such plugin. Prefer plugins from established developers with real security track records.
Conclusion
The credit lines in WordPress 7.1.3 mark a permanent change: AI WordPress flaws are no longer a research curiosity but the main engine of vulnerability discovery. Anthropic, OpenAI’s research partners, and autonomous testing firms now find the bugs that once took human researchers months — at $25 and ten hours a pop. That is excellent news for the security of the software itself, and genuinely bad news for anyone who patches slowly. The 450-report July, the $90,000 bounty quarter, Google’s paused program, and the five-hour disclosure-to-exploitation median all point the same way: the window between “vulnerability published” and “vulnerability exploited” has collapsed. Site owners cannot control how fast flaws are found anymore — but they can control how fast they patch. Automatic updates, a firewall, fewer plugins, and real monitoring are no longer best practices. They are the minimum.
If keeping up with this pace sounds like a full-time job, that is because it nearly is. Qamar Digital Ventures helps businesses stay secure and grow online. Our website care plans handle updates, backups, and security monitoring for you — so AI-discovered flaws get patched before they become your problem. Visit https://qamardigitalventuresllc.com/ to learn more, or contact us to discuss your project.
Sources
- The Repository — WordPress 7.1.3 Patches Seven Vulnerabilities as Bug Bounty Payouts Top $90,000: https://www.therepository.email/wordpress-7-1-3-patches-seven-vulnerabilities-as-bug-bounty-payouts-top-90000-in-90-days
- Coastal Digital — The Bots Found 450 Ways Into WordPress Last Month: https://coastaldigital.org/blog/the-bots-found-450-ways-into-wordpress-last-month/
- WP Poland — AI and WordPress Vulnerabilities: Who Pays for the Patches: https://wppoland.com/en/ai-vulnerability-reports-flood-wordpress-who-pays-for-patches/
- RS Web Solutions — AI Detects WordPress Vulnerabilities, Urges Fast Fix: https://www.rswebsols.com/news/ai-quickly-identifies-wordpress-vulnerabilities-giving-new-zealand-websites-only-hours-to-fix-not-weeks/
