📥 Hello, and greetings from the Central Office!
We like to be upfront about what is still in progress. A few items carried over from May: the Beehiiv subscriber sync issue, remaining WPCS remediation work, WP Engine compatibility investigations, and an SSL renewal for passwordprotectwp.com. All are actively being worked on.
For June, our focus is on completing the Protect WP Videos release, continuing security hardening, resolving the Beehiiv sync, launching updated onboarding flows, and expanding our cross-sell campaigns between PDA and PPWP users.
Thank you for following along this month. Sharing our work transparently is something we genuinely care about, and your continued support makes it all worthwhile. 💙
🎥 Did you know we have a YouTube channel focused on WordPress security?
We share step-by-step tutorials on how to secure your site and protect your content.
⚡️ Watch our latest video:
How to Create a Free Coming Soon Page in WordPress
👉 Explore the full channel:
youtube/@wpdefenselab
Don’t miss the next one 🚀
Welcome to WP Defense Lab — your weekly brief on the WordPress news that impacts your business.
This week: over 100 copyright suits challenge AI companies' fair use defense for training data, while a licensing market worth billions proceeds without most creators. The Copyright Alliance documents why the fourth fair use factor will be decisive in court.
In WP Radar: Kinsta's data on WordPress bot server costs, WordPress's 24-hour plugin update delay pushback, a Munich court holding Google liable for AI Overview errors, and TheSoul Group's 60-platform streaming reach.
AI Companies Built a Licensing Market. Most Creators Aren't in It.
The Copyright Alliance tracks over 100 active copyright suits where AI companies invoke fair use to avoid paying for training data, while a real, growing licensing market operates without most independent creators.
The Copyright Alliance tracks over 100 active copyright lawsuits against AI companies.
The common thread across all of them is fair use. AI companies argue that training their systems on copyrighted works is "transformative," a legal framing that, in their version, eliminates the need to license or pay for the content they ingested.
That argument has a structural problem that courts are beginning to work through.
The Factor That Controls the Analysis
Fair use analysis weighs four factors. Courts have treated the fourth, the effect of the use upon the potential market for or value of the copyrighted work, as the controlling factor since Harper & Row v. Nation Enterprises (1985).
The Supreme Court reaffirmed that hierarchy in Warhol v. Goldsmith: when a new work's purpose is highly similar to the original, it substitutes for it in the market.
That substitution is the problem. An AI system trained on a publisher's articles generates outputs that compete directly with those articles.
"Transformative use" is one element of the first factor. It is not a standalone argument, and it does not override the fourth factor when market harm exists.
Evidence has also emerged in discovery that some AI models reproduce copyrighted material verbatim in their outputs, which collapses the transformation argument entirely.
The Licensing Market That Already Exists
The persistent claim from AI companies is that the market for AI training data is speculative or underdeveloped.
That framing is false. AI companies of all sizes are entering licensing agreements with publishers, news organizations, record labels, and movie studios.
Smaller AI companies, including Bria.ai, Moonvalley, ProRata, and KLAY Vision, have built their businesses entirely on licensed content. They are not doing this as an ideological statement.
Licensed, professionally produced content produces better, more reliable AI systems than datasets assembled from shadow library websites.
The existence of this market matters legally. Courts evaluate fair use against actual markets AND markets likely to develop.
An established, growing licensing market means companies ingesting content without permission are bypassing a real commercial transaction.
Who Bears the Cost of "Free" Training Data
When large AI companies claim fair use to avoid licensing costs, they impose that cost on two groups. Creators and publishers lose compensation from a market that exists specifically to pay them.
Smaller AI companies that chose to license training data compete against companies that did not, at a structural cost disadvantage. Ethical behavior becomes an economic penalty.
Courts allowing broad fair use claims for AI training would not just harm individual rights holders. It would drive the licensed AI model out of business and consolidate the market around companies with the resources to absorb litigation risk.
What This Means for WordPress Publishing Operations
Your content may already be in someone's training dataset.
Whether you are in the licensed side of that market depends on what agreements exist and what your content archive demonstrates.
The AI training licensing market rewards publishers who can document their output, prove its quality, and negotiate independently of any platform that currently distributes their work.
Owned infrastructure is the prerequisite for any licensing conversation, not because WordPress itself gets you into the licensing market, but because publishers who own their archives operate from a position with something specific to offer.
Publishers dependent on a third-party platform for distribution are negotiating from a position where the platform has already extracted the first layer of value.
The Practical Question
The question is not whether to block AI crawlers.
It is whether you are positioned to participate in the revenue tier that already exists.
Half your market is one app away.
Your business is already on Instagram, SMS, and web chat. But 52 million immigrants in the US rely on WhatsApp to connect with businesses they trust — not email, not phone calls.
Wati helps you show up on WhatsApp and every channel they use. Are you still not there?
This week in WP Radar: WordPress sites absorbing AI bot server load, the 24-hour plugin update delay that applies to manual installs too, Google's new AI liability exposure in European courts, a content company running 60-plus streaming platforms, and how brand AI visibility budgets are shifting before the ad market matures.
AI Bot Traffic Becomes a WordPress Infrastructure Tax. Kinsta analyzed 10 billion requests across its managed WordPress infrastructure and found a single bot (ClaudeBot) generated 3.75 million add-to-cart requests within a 24-hour window, each one triggering PHP execution and a database query on a WooCommerce store's dynamic endpoints. A second misbehaving crawler loop generated 550 million requests over 30 days on one site. WordPress operators who haven't reviewed bot access rules are absorbing that server load without any traffic or revenue return.
WordPress Plugin Delay Applies to Manual Updates Too. WordPress.org's Protect the Shire initiative introduced a 24-hour delay before updates propagate through the update system. Plugin developers pushed back on a detail that wasn't clear in the original announcement: the delay applies to all updates through WordPress's update system, not just automatic ones. Manual updates from the dashboard carry the same 24-hour wait. For sites running security-critical plugins, the gap between a disclosed vulnerability and an available patch now has an additional day built in.
Munich Court Holds Google Liable for AI Overview Errors. The Regional Court of Munich issued a temporary injunction on May 28, 2026, barring Google from repeating false AI Overview claims that tied two local publishers to scams. The court treated the AI Overview as Google's own authored content, ruling it produces "independent, new, and substantive statements" that carry liability. For WordPress publishers, structured entity data on owned infrastructure is now the foundation for any dispute over how AI systems represent their brand.
5-Minute Crafts Parent Operates 60-Plus Streaming Platforms. TheSoul Group built 5-Minute Crafts into a brand with 400-plus million followers across social platforms, then converted that momentum into 8,000 long-form episodes distributed across 60-plus streaming platforms in 180 countries. Its white-label division manages distribution for Crayola, Hasbro, Mattel, and the Harlem Globetrotters. The model shows what a content operation with owned distribution infrastructure looks like: platform algorithms drive discovery, but the content library and streaming relationships belong to the operator.
Anua's earned media value grew from $665,000 in 2020 to $290.5 million in 2025, a 437x increase driven by systematic creator investment that expanded the brand's creator network from 142 to 23,900 contributors over six years.
The 437x EMV growth came from systematic creator investment, not a single viral moment. Anua's inflection point was Amazon entry in 2022, giving creator-led momentum a retail conversion path. Independent publishers building owned subscriber lists are working a different compounding model, one where the audience data and direct relationship accumulate as assets on infrastructure the publisher controls.
That’s all for this week!
Michael - Operator @WP Folio - now WP Defense Lab. Same Plugins. Different Name.